https://orcid.org/0000-0001-6100-5452
Figures
Abstract
Background
Stem cell research, particularly in the domain of induced pluripotent stem cell (iPSC) technology, has shown significant progress. The integration of artificial intelligence (AI), especially machine learning (ML) and deep learning (DL), has played a pivotal role in refining iPSC classification, monitoring cell functionality, and conducting genetic analysis. These enhancements are broadening the applications of iPSC technology in disease modelling, drug screening, and regenerative medicine. This review aims to explore the role of AI in the advancement of iPSC research.
Methods
In December 2023, data were collected from three electronic databases (PubMed, Web of Science, and Science Direct) to investigate the application of AI technology in iPSC processing.
Results
This systematic scoping review encompassed 79 studies that met the inclusion criteria. The number of research studies in this area has increased over time, with the United States emerging as a leading contributor in this field. AI technologies have been diversely applied in iPSC technology, encompassing the classification of cell types, assessment of disease-specific phenotypes in iPSC-derived cells, and the facilitation of drug screening using iPSC. The precision of AI methodologies has improved significantly in recent years, creating a foundation for future advancements in iPSC-based technologies.
Conclusions
Our review offers insights into the role of AI in regenerative and personalized medicine, highlighting both challenges and opportunities. Although still in its early stages, AI technologies show significant promise in advancing our understanding of disease progression and development, paving the way for future clinical applications.
Citation: Vo QD, Saito Y, Ida T, Nakamura K, Yuasa S (2024) The use of artificial intelligence in induced pluripotent stem cell-based technology over 10-year period: A systematic scoping review. PLoS ONE 19(5): e0302537. https://doi.org/10.1371/journal.pone.0302537
Editor: Li-Ping Liu, Affiliated Hospital of Jiangsu University, CHINA
Received: February 2, 2024; Accepted: April 9, 2024; Published: May 21, 2024
Copyright: © 2024 Vo et al. This is an open access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited.
Data Availability: All relevant data are within the manuscript and its Supporting Information files.
Funding: The author(s) received no specific funding for this work.
Competing interests: The authors have declared that no competing interests exist.
Abbreviations: AI, artificial intelligence; aiPSC, artificially induced pluripotent stem cell; ALS, amyotrophic lateral sclerosis; AUC, area under the curve; CM, cognitive maps; CNN, convolutional neural network; CT, Classification Tree; DAGSVM, Directed Acyclic Graph; DL, Deep learning; ECC, embryonal carcinoma cell; ESC, embryonic stem cells; hiPSC, Human induced pluripotent stem cells; ICC, intraclass correlation coefficient; IOPD, Infantile onset Pompe disease; iPSC, Induced pluripotent stem cells; k-NN, k-Nearest Neighbour; lncRNA, long non-coding RNA; MEA, microelectrode array; ML, Machine learning; MPI, magnetic particle imaging; NB, naïve Bayes; PRISMA, Preferred Reporting Items for Systematic Reviews and Meta-Analysis; PSC, pluripotent stem cell; RNN, recurrent neural network; ROC, Receiver Operating Characteristic; SVM, Support Vector Machines; wndchrm, compound hierarchy of algorithms representing morphology
Introduction
Stem cell research has advanced significantly since 1961, when bone marrow-derived multipotent stem cells were first identified [1]. Stem cells are unique cells with the capability to continually replicate through mitosis, leading to the formation of more cells. This process generates two distinct cell types: one that evolves into a specific cell type and another that retains the capability of self-renewal [2]. Stem cells are broadly classified into three types: induced pluripotent stem cell (iPSC), embryonic stem cell (ESC), and adult stem cell (ASC) [3]. The iPSC and ESC are categorized as pluripotent stem cell (PSC) due to their capacity to transform into the three germ layers: ectoderm, mesoderm, and endoderm. In 2006, Kazutoshi Takahashi and Shinya Yamanaka successfully transformed mouse somatic cells into iPSC by introducing specific transcription factors known as Oct4, Sox2, Klf4, and c-Myc using a viral vector [4]. Following this, various methods have been used to reprogram different types of mouse and human somatic cells into iPSC [5]. This innovative method of reprogramming human cells has sparked immense interest in both scientific and medical fields. iPSC provides an alternative to human ESC as a source of pluripotent cells. A significant advantage of induced pluripotent stem cells is that they are derived from somatic cells that could be obtained non-invasively. These cells carry the individual’s genetic characteristics, which can reduce the risk of immune rejection [6].
The attention to iPSC-based therapies is increasing in the field of modern medicine. Their application in disease modelling, drug screening, and regenerative medicine is expanding exponentially [7]. The iPSC is pivotal in disease modelling due to its self-renewal capability and ability to differentiate into all human body cell types. This makes them ideal for creating various disease models for research [8–10]. Patient-specific iPSC is particularly valuable in developing targeted therapeutic strategies and drug development. Furthermore, iPSC from both normal and diseased cells can be differentiated into neurons, hepatocytes, cardiomyocytes, etc., for evaluating toxicity and side effects, which are critical factors in the development of therapeutic molecules [11]. In regenerative medicine, iPSC is being used to repair or regenerate damaged or degenerated tissues. This is achieved by creating the organ tissues in laboratories from iPSC and transplanting them to the injured area. This therapy holds promise for treating conditions such as hematopoietic disorders, musculoskeletal injuries, spinal cord injuries, and liver damage [12–14].
Various techniques have been developed for creating iPSC, such as using retroviruses or lentiviruses for gene transduction and chemical induction. However, the process of generating iPSC is typically slow and not very efficient, taking about 1–2 weeks for rodent cells and 3–4 weeks for human cells, with generally low success rates. Moreover, evaluating the quality of iPSC by examining colony morphology is prone to human error, presenting a significant challenge that must be addressed before pursuing further experimental or therapeutic uses. Despite the advancements in enhancing both the efficiency and speed of iPSC cultivation, this process remains resource-consuming, necessitating the development of an automated system to minimize errors and enhance iPSC analysis. Recently, artificial intelligence (AI) technologies, including machine learning (ML) and deep learning (DL), have been employed to enhance regenerative therapy. The implementation of these AI-driven approaches could refine the management of clinical trials for innovative stem cell therapies designed for a multitude of diseases. AI contributes to the personalization of treatment protocols for individual patients, the forecasting of clinical outcomes, and the practical organization of patient recruitment. These advancements have the potential to diminish the challenges inherent in these trials and reduce the overall expenses [15].
AI technology involves the development of computer systems designed to perform tasks traditionally requiring human intelligence, encompassing learning, reasoning, perception, and problem-solving. The objective of AI systems is to mimic human cognitive abilities, functioning autonomously and continuously improving their capabilities by learning from data and past experiences [16]. The concept of AI has existed for many years, but recent advancements in ML and DL have enabled the development of more complex AI systems. Nowadays, AI systems can process vast datasets to predict outcomes, categorize objects, and execute other intricate functions.
Machine learning represents a sophisticated and advancing technology that enables computers to identify and categorize patterns in extensive datasets without explicit programming. It is an interdisciplinary field, intertwining computer science, mathematics, philosophy, control theory, determinism, and other areas [17]. This approach focuses on replicating or emulating human learning processes using computers. These techniques allow researchers to filter through extensive datasets, identify patterns, make predictions, learn from errors, and adapt their methods for improvement without explicit programming [18].
Deep learning is a specialized branch of ML which employs artificial neural networks, and some other algorithms, for data processing. These networks are structured to resemble the human brain, enabling the identification of more intricate patterns and decision-making based on trained data [19]. This method has significantly transformed the AI landscape, equipping machines to undertake tasks previously considered unachievable. A notable strength of DL lies in its capacity to manage extensive and complex datasets. Deep learning algorithms can process millions of data points and discern patterns beyond human recognition [20]. Furthermore, DL distinguishes itself with its adaptive learning capabilities over time, whereas traditional ML methods often lack a memory component and require manual tuning [21,22]. This self-optimization allows DL algorithms to progressively enhance their performance as they are exposed to more data. Among these algorithms, the convolutional neural network (CNN) is predominantly utilized for image classification tasks [23].
AI has been instrumental in advancing iPSC technology, especially through non-invasive cell identification, genomic and proteomic data analysis, and the enhancement of targeted therapies [24] (Fig 1). The ongoing advancements in AI hold a promise to radically transform medical science. However, given the rapid pace of this field, it is imperative to conduct extensive research and validation to leverage the full potential of AI in iPSC technology. This review aims to investigate and illustrate a comprehensive review of existing studies on how AI-based methodologies contribute to the advancement of iPSC technology.
Materials and methods
This systematic review was conducted adhering to the 2020 Preferred Reporting Items for Systematic Reviews and Meta-Analysis (PRISMA) guidelines and principles [25].
Searching strategy
In December 2023, a primary search was conducted across three electronic databases: PubMed, Web of Science, and Science Direct. The search utilized terms and their synonyms, including “artificial intelligence,” “deep learning,” “machine learning,” and “induced pluripotent stem cell,” to identify potential articles in the targeted databases using the specified search terms. After this initial step, duplicate entries were removed. The next step involved a preliminary screening of the remaining papers by reviewing their titles and abstracts. This was followed by a thorough full-text review of each paper, conducted independently by two investigators to determine their relevance and eligibility. A third investigator was brought in to provide a definitive judgment if there were any disagreements or uncertainties.
Additionally, to ensure comprehensive coverage, a manual citation search was also undertaken. This involved screening the reference lists of all retrieved studies to uncover any other pertinent research that might have been missed in the initial database search.
Inclusion and exclusion criteria
This study included quantitative research that applied AI-based imaging analysis to iPSC from either animals or humans. No language or publication date limitations were imposed, given the emerging nature of this field. Exclusions were made for preprints, conference papers, qualitative studies, and quantitative studies focusing on AI technology for cell types other than iPSC. Studies were also excluded if the full text was not available.
Quality assessment
All studies meeting the inclusion criteria were evaluated for the best practice in AI research according to DOME guideline [26].
Data extraction
Data from the included studies were independently extracted and assessed by two reviewers. If there were any disagreements, a third reviewer would be consulted. Quality assessment techniques were applied to ensure that studies met the inclusion criteria. Extracted data encompassed bibliographic details (authors, publication year, funding, title, language, journal), and details of the iPSC line, AI technique, and the accuracy of the AI algorithm were also collected.
Results
Search result
In general, from the initial 872 articles retrieved from three electronic databases, 440 were removed after deduplication. Following this, 432 studies underwent screening of titles and abstracts based on the inclusion criteria. After this initial screening, 95 articles were selected for full-text analysis. However, 16 studies were excluded after a detailed review, resulting in 79 studies being included in this systematic review (Fig 2).
Characteristics of included studies
Of the 79 studies, one focused on animal iPSC, 75 on human iPSC, and three utilized artificially created iPSC. Additional details about these studies can be found in S1 Table in S1 File.
The demographic analysis highlighted the United States as the predominant contributor in this field, accounting for 31 of the 79 studies (39.2%). Europe and Japan followed, each contributing 15 studies (18.9%). Other regions, including China, the United Kingdom, Taiwan, Korea, and Canada, had a lower research volume, with contributions ranging from 2 to 4 studies per region. Despite year-to-year variance, there is an increasing trend of publication volume regarding the application of AI in the iPSC field over the years, peaking in 2022 with twenty studies published (Fig 3).
Applications of AI in iPSC
The applications of AI in iPSC field can be classified into 5 major categories (Fig 4).
Cell culturing and processing
A total of 44 articles were identified that leveraged AI technologies to enhance cell processing methods. Among these, 23 studies employed AI for the classification of various cell types based on morphological images, 11 aimed at assessing the biological attributes of cells, five focused on managing the quality of the cell culture process, two utilized AI for monitoring cells after transplantation, and three developed artificial iPSC using existing datasets. Data inputs varied across these studies, with 39 employing cell imagery, two using biological data, and three relying on genetic data (Table 1).
Genetic analysis
In genetic analysis, ten articles employed AI technologies. Of these, seven studies applied AI for RNA sequencing analysis, two for DNA analysis, and one for evaluating epigenomic characteristics (Table 2).
Cell function monitoring
Eight studies applied AI for monitoring cell functions, employing various inputs: six studies analysed calcium transient signals, one study focused on action potential signals, and one used microelectrode array (MEA) data (Table 3).
Disease model
Four studies implemented AI to distinguish diseased cells from healthy controls. This included one study that analysed images of Huntington iPSC to identify diseased colonies based on cell morphology [89]. one study that evaluated iPSC-derived motor neurons from individuals with and without amyotrophic lateral sclerosis (ALS), [90] and two studies that assessed contracting signals to detect abnormal cardiomyocytes [91,92] (Table 4).
Drug screening model
Thirteen studies applied AI in the context of drug modelling, with eight focusing on identifying the effects of pharmaceuticals and the mechanisms of action of drugs, and the remaining five studies assessed drug-induced toxicity on iPSC (Table 5).
Main AI algorithms employed in iPSC research
Regarding the application of AI algorithms, convolutional neural network was used in 33 research studies, comprising 41% of the total analysis. Machine learning methodologies were adopted in 44 studies, making up 55.6% of the comprehensive research effort. Among these methodologies, the most frequently used techniques included support vector machine (SVM), which was featured in 21 studies (26.5%), and random forest model, which was employed in 16 studies (20.2%). Other algorithms such as K-Nearest Neighbor, Cognitive Map,… were reported in fewer than 10% of the studies.
Additional information of included studies is provided in S1 Table in S1 File.
Quality of included studies
Of the 79 studies, data and computer codes were accessible in 29% of the cases, and only two studies provided best practice reports [29,97]. However, 85% of the studies provided detailed descriptions of the employed AI methodologies, while 87% reported the performance metrics. Furthermore, the practices of data splitting and hyperparameter tuning were reported in 77% and 67% of the studies, respectively. Detailed information can be found in S2 Table in S1 File.
Discussion
Over the past twenty years, induced pluripotent stem cell (iPSC) technology has undergone significant advancements and is now being explored for its potential in regenerative medicine. The iPSC therapy holds the promise of creating individualized treatments, avoiding the ethical concerns associated with embryonic stem cells and the risk of immune rejection. However, the adaptation of iPSC technology for clinical applications is facing significant hurdles, including low efficiency and considerable variability in iPSC reprogramming and differentiation methods [106,107], as well as the emergence of undifferentiated and teratoma-induced phenotypes [108]. The integration of artificial intelligence (AI) has significantly advanced the field of iPSC research. This advancement is evident in various aspects, including the classification of iPSC colonies, the non-invasive differentiation between normal and abnormal cells, and the analysis of cellular morphology. Furthermore, AI has been implemented in other areas of iPSC-based technology, encompassing drug testing, disease modelling, and regenerative treatments [109].
The evolution of AI applications in iPSC research
The evolution of AI applications in iPSC technology over the last decade highlights a transformative journey from simplicity to complexity, both in terms of data and analytical methods. In the early stages, AI is predominantly used to analyse cell image datasets [31,38,43]. These initial studies laid the groundwork for using machine learning to automate tasks like cell classification and tracking, employing relatively straightforward algorithms such as support vector machine (SVM) [44,54], and random forest [56]. These algorithms were adept at parsing visual data, aiding researchers in achieving a significant understanding of cell morphology and behaviour through images.
In the early 2020s, iPSC research witnessed a shift towards incorporating more complex and varied types of data, including genetic information [60] and biological data [30,104]. This transition was facilitated by the broader availability of high-throughput sequencing technologies and advanced imaging techniques, which produced richer datasets capturing the vast complexities of cellular biology. The applications of AI expanded in scope and ambition, moving from basic image recognition tasks to more sophisticated analyses like the modelling of disease processes and the examination of iPSC clonal expansion. This period also saw an increased integration of advanced machine learning algorithms, particularly deep learning models such as convolutional neural network (CNN), which were able to handle the intricacies of large and complex datasets.
The integration of AI technology in iPSC research not only aimed at enhancing the understanding of cellular processes but also at tackling emerging challenges in regenerative medicine and disease modelling [24]. The advancements in computational power and algorithmic sophistication allowed researchers to delve deeper into the biological intricacies of iPSC, paving the way for groundbreaking applications in disease modelling Witmer [90,92] and drug discovery [97,103]. This evolution from simpler machine learning techniques to advanced deep learning approaches mirrors the growing complexity and scale of biological data, highlighting a trend towards increasingly ambitious and nuanced scientific inquiries in the field of iPSC technology.
Current applications of AI in iPSC research
Within the field of induced pluripotent stem cells (iPSC), a significant challenge arises from genomic instability initiated by the reprogramming process [110] and variations in the culture system [111]. These factors can affect the differentiation capabilities of iPSC, consequently influencing their therapeutic potential. Also, iPSC derivatives sometimes remain in the fetal stage of development, which may reduce the efficacy of iPSC-based treatments. To ensure the safety and effectiveness of iPSC therapy, it is crucial to continuously evaluate the cells derived from iPSC at different stages of their development. Currently, this assessment mainly depends on the judgment of experienced cell culture specialists. These experts often assess iPSC development and maturity by examining changes in cell shape and the expression of specific cellular markers. This method is not only time-consuming but also susceptible to personal bias. Relying solely on manual techniques for cell quality evaluation is not feasible for mass-producing therapeutic cells [112]. AI technology has consistently demonstrated its effectiveness in assisting the cultivation and maintenance of iPSC, particularly in identifying cell colonies and their functions. Since the pioneering work by Henry Joutsijoki et al. in 2014, which used ML to evaluate the quality and categorize iPSC colonies [43], many studies have applied a wide range of AI algorithms for analysing differentiation levels and morphological alterations within iPSC cultures. With the enhancements in accuracy, AI algorithms have established a reliable framework for various applications in iPSC research, including the accurate classification of iPSC colonies, identification of cellular morphology, non-invasive characterization of cell therapies, distinguishing between healthy and unhealthy cells, and recognition of previously unknown morphological traits [38,46,65]. The advantage of AI-based methodology lies in its independence from labelling, genetic alterations, or immunostaining, making it applicable to various scenarios that require the use of intact, live cells. Recently, AI technology has been employed to construct automated systems capable of high-throughput screening to reliably verify cells’ identity and function throughout the entire production process. Furthermore, the integration of AI and robotic technology has led to the development of automated culture systems such as the CompacTSelecT (CTST) [113]. These advancements allow scientists to remotely implement protocols, facilitate the automatic maintenance and differentiation of iPSC into various cell types, and conduct highly efficient high-throughput screening using iPSC. This automation not only enhances the efficiency and consistency of cell culture workflows but also minimizes human error, leading to more reliable and reproducible results in stem cell research and applications.
For disease modelling, network-based screening using AI technologies is a powerful method for identifying molecules that address disrupted gene networks in human diseases, offering promising candidate molecules for in vivo validation [71,73] Furthermore, AI now extends its application beyond the generation of differentiated cells from iPSC. By integrating AI into the culture and maintenance of iPSC, it is possible to create complex tissue structures containing multiple cell types, essential for the development of organoids. These multicellular formations, illustrating three-dimensional tissue architecture, provide deeper insights into disease mechanisms compared to simpler cellular models. Recent studies highlight the significance of AI in refining the structure and functionality of organoids as models for disease [51,80]. Moreover, introduced in 2019, the DeepNEU platform, which integrates recurrent neural network (RNN), cognitive mapping (CM), and evolutionary systems, can exploit data from existing databases regarding essential genes and proteins that regulate and sustain signalling pathways in hiPSC and hESC. This platform produces simulations of artificially induced pluripotent stem cell (aiPSC) that are consistent with the outcomes observed for actual iPSC [70]. This model has been successfully applied in generating aiPSC for infantile-onset Pompe disease (IOPD) and skeletal muscle, demonstrating high accuracy in replicating gene and protein expression and phenotypic characteristics [60]. These advancements have significant implications for the progression of disease research and clinical practices, highlighting the potential of AI-driven technologies in enhancing disease modelling, prototyping, and pathological prediction. This approach is particularly valuable for both common and rare diseases, offering a cost-effective alternative to traditional methods.
In drug discovery and development, AI is becoming increasingly integrated with iPSC technology, paving the way for novel, patient-specific assessment methods and the identification of new therapeutic agents for treating diseases. The synergy between AI and iPSC research extends beyond mere image analysis to the analysis of genomic data and pathological conditions [71,78,114]. These advancements open avenues for efficient selection, analysing the interactions and effectiveness of potential drug candidates through in silico virtual screening techniques, thereby revolutionizing drug discovery research with iPSC [115]. The initial research in this field was conducted by Eugene K. Lee et al. and Heylman Christopher in 2015. In these studies, iPSC-derived cardiomyocytes were used to evaluate drug-induced cardiotoxicity. Machine learning was employed to distinguish between normal and abnormal contractions [96] and membrane depolarization voltage following exposure to cardioactive drugs [95]. This approach showed an effective method in assessing the cardiotoxicity of various compounds, achieving an accuracy rate exceeding 80%. Subsequently, numerous studies have demonstrated the success of computational methods in predicting treatment responses [103] and in identifying internal changes post-treatment with certain substances [116] and pharmaceutical toxicity in iPSC and organoid models [94,104]. Currently, AI technology has been employed to develop high-throughput, label-free drug screening systems. AI-driven systems such as Deep-SeSMo and SSGraphCPI have demonstrated their potential in identifying new therapeutic agents to treat various diseases [117,118]. In 2020, Mahnaz Maddah et al. introduced PhenoTox, an 18-layer convolutional neural network system capable of detecting drug-induced structural changes in live hiPSC-hepatocytes and hiPSC-cardiomyocytes from brightfield microscopy images before these alterations are observable by humans. [119]. By 2022, Manuela Jaklin had developed the TeraTox system, a machine-learning algorithm with the ability to examine concentration-dependent cytotoxicity and changes in genetic expression induced by pharmaceutical compounds in hiPSC-derived embryoid bodies [100]. Particularly, during the COVID-19 pandemic in 2020, the DeepNEU was deployed for modelling lung aiPSC infection with SARS-CoV-2, facilitating the rapidly identification of antiviral therapeutic targets and opportunities for drug repurposing [61]. These advancements underscore AI’s transformative role in drug development, highlighting its ability to deliver innovative and effective solutions in healthcare. The integration of AI and stem cell research not only expedites drug discovery processes but also paves the way for personalized medicine and targeted treatment strategies.
Limitation and quality considerations in AI applications
The susceptibility of ML and DL algorithms to artifacts, along with the inherent biological complexity of cell-based assays, introduces numerous potential sources of noise and variability [120]. These challenges cover a broad spectrum of issues, including inconsistencies in cell plating, variations in cell growth and response, differential responses across plates, inter-plate variability, changes across different experimental runs, and edge effects—where conditions at the edges of the plate are different from those at the center. In addition, clonal effects may add variability that complicates the analysis of results from cellular assays. Several strategies can be employed to address these challenges. Standardizing cell plating protocols, growth conditions, and assay procedures can markedly decrease variability. The adoption of automation and the use of high-throughput screening techniques can further standardize experimental workflows, reducing human error and inconsistency. Furthermore, the employment of advanced imaging and data analysis tools enables a more precise capture and analysis of biological responses, providing clearer insights into the data. Through these approaches, the reliability and accuracy of ML and DL applications in cell-based research can be significantly enhanced, paving the way for more robust scientific discoveries.
Moreover, the distinction between biological and technical replicates is another critical consideration. Algorithms must be trained and tested on data that accurately reflect both the biological variability and the technical reproducibility of the experiment. Failure to differentiate between these factors can lead to overfitting, where the model performs well on the training data but poorly on unseen data. To address this, it is essential to design experiments to ensure that the data used for training and testing the algorithms genuinely represent the diversity and complexity of biological systems. Furthermore, developing rigorous validation and cross-validation methods is important to verify the robustness and generalizability of AI models in cell-based research [121].
Another obstacle to integrating AI technologies into biological research is the difficulty in achieving reproducibility. This challenge stems from several factors, including the limited availability of the original computer codes and datasets, inadequate descriptions of the methodologies used, and the partial disclosure of findings [122]. Our analysis highlights that the code utilized was available in 29% of the cases, and only 3% of studies provided in best practice reports [29,97]. Additionally, the application of AI in biology is characterized by its diverse empirical methods, which, although may be successful in certain conditions and in specific laboratory environments, often do not have robust theoretical validation [123].
Implications for future applications
AI holds substantial potential to revolutionize and accelerate the advancement of various aspects of iPSC technology, from the cell culturing process and drug discovery to disease modelling and the development of cellular therapies. Capable of extracting valuable insights from vast amounts of molecular and genomic data—tasks would be unfeasible for humans—AI could significantly advance iPSC research and development. Despite the potential benefits of AI in enhancing iPSC research, substantial technical hurdles must be overcome before its applications can be widely employed. Consequently, extensive research is required to address these barriers and fully harness the potential of AI in iPSC technology.
As advancements in AI technologies progress and the availability of high-quality data expands, the potential to refine and customize AI algorithms for application in regenerative medicine is expanding. Innovations in AI, computer vision, and robotics hold the promise of discovering new insights that could lead to transformative developments in iPSC technology. The integration of AI with other nanotechnology, genome editing, and 3D bioprinting could pave the way for breakthroughs in developing personalized regenerative therapies. By fostering interdisciplinary cooperation, ensuring the ethical development and application of these technologies, it is possible to fully leverage AI’s potential to develop customized and effective regenerative treatments.
Finally, it is critical to establish robust benchmarking standards and criteria for the development and assessment of AI algorithms. These ground truth datasets, characterized by their accurate labelling and broad acceptance, enable an unbiased evaluation of an algorithm’s effectiveness. They serve as a cornerstone for comparing various models and methodologies and aiding in the discernment of the most efficient techniques for specific applications. Furthermore, standardized reporting protocols are crucial for significantly improving the quality of research in AI applications.
Conclusions
Artificial intelligence technology, particularly ML and DL, offers substantial improvements over human capabilities in terms of accuracy, speed, and cost-efficiency in data analysis. These methods not only assist researchers but also pave the way for the transformation of decision-making processes in the biomedical field. AI technology has been significantly impacting every stage of stem cell studies, from laboratory research to clinical application. Its potential to revolutionize iPSC manufacturing lies in offering cost-effective, rapid, and robust screening methods for a large number of iPSC lines and their derivatives, essential for obtaining cells suitable for clinical use. Additionally, AI methods are increasingly being utilized in iPSC-based drug discovery, aiding drug efficacy, toxicity, and pharmacokinetics predictions.
The number of research studies in this area is constantly growing, pushing the boundaries of what AI can achieve in biomedical contexts. However, the implementation of these methods in clinical settings still requires substantial groundwork. A significant challenge lies in the need for extensive volumes of carefully curated, structured training data, which is crucial for generating meaningful results. Researchers must pay attention to factors that can influence the outcomes of AI applications, including the quality of data in the training set, the variability in sample size and unexpected events that algorithms may not predict.
Limitations
Our study has certain limitations. Firstly, our research focuses on the application of AI in the context of iPSC therapy, we do not provide an in-depth analysis of the specific AI algorithms used in each referenced study. Second, there is inconsistency in reporting AI’s accuracy, and, in some instances, it is missing. This inconsistency poses challenges in evaluating the efficacy of AI within iPSC therapy. Therefore, further detailed investigation and refinement of AI applications in iPSC therapy are necessary to fully understand and utilise their true therapeutic potential.
Supporting information
S1 Checklist. Preferred Reporting Items for Systematic reviews and Meta-Analyses extension for Scoping Reviews (PRISMA-ScR) checklist.
https://doi.org/10.1371/journal.pone.0302537.s001
(PDF)
References
- 1. Till J.E. and Mc C.E., A direct measurement of the radiation sensitivity of normal mouse bone marrow cells. Radiat Res, 1961. 14: p. 213–22. pmid:13776896
- 2. Ramakrishna R.R., et al., Stem cell imaging through convolutional neural networks: current issues and future directions in artificial intelligence technology. PeerJ, 2020. 8: p. e10346. pmid:33240655
- 3. Cable J., et al., Adult stem cells and regenerative medicine-a symposium report. Ann N Y Acad Sci, 2020. 1462(1): p. 27–36. pmid:31655007
- 4. Takahashi K. and Yamanaka S., Induction of pluripotent stem cells from mouse embryonic and adult fibroblast cultures by defined factors. Cell, 2006. 126(4): p. 663–76. pmid:16904174
- 5. Ho R., Chronis C., and Plath K., Mechanistic insights into reprogramming to induced pluripotency. J Cell Physiol, 2011. 226(4): p. 868–78. pmid:20945378
- 6. Bellin M., et al., Induced pluripotent stem cells: the new patient? Nat Rev Mol Cell Biol, 2012. 13(11): p. 713–26. pmid:23034453
- 7. Rami F., et al., Recent Advances in Therapeutic Applications of Induced Pluripotent Stem Cells. Cell Reprogram, 2017. 19(2): p. 65–74. pmid:28266864
- 8. Lee G., et al., Modelling pathogenesis and treatment of familial dysautonomia using patient-specific iPSCs. Nature, 2009. 461(7262): p. 402–6. pmid:19693009
- 9. Moad M., et al., A novel model of urinary tract differentiation, tissue regeneration, and disease: reprogramming human prostate and bladder cells into induced pluripotent stem cells. Eur Urol, 2013. 64(5): p. 753–61. pmid:23582880
- 10. Yazawa M., et al., Using induced pluripotent stem cells to investigate cardiac phenotypes in Timothy syndrome. Nature, 2011. 471(7337): p. 230–4. pmid:21307850
- 11. Gunaseeli I., et al., Induced pluripotent stem cells as a model for accelerated patient- and disease-specific drug discovery. Curr Med Chem, 2010. 17(8): p. 759–66. pmid:20088756
- 12. Liu H., et al., In vivo liver regeneration potential of human induced pluripotent stem cells from diverse origins. Sci Transl Med, 2011. 3(82): p. 82ra39. pmid:21562231
- 13. Nori S., et al., Grafted human-induced pluripotent stem-cell-derived neurospheres promote motor functional recovery after spinal cord injury in mice. Proc Natl Acad Sci U S A, 2011. 108(40): p. 16825–30. pmid:21949375
- 14. Suzuki N., et al., Generation of engraftable hematopoietic stem cells from induced pluripotent stem cells by way of teratoma formation. Mol Ther, 2013. 21(7): p. 1424–31. pmid:23670574
- 15. Sniecinski I. and Seghatchian J., Artificial intelligence: A joint narrative on potential use in pediatric stem and immune cell therapies and regenerative medicine. Transfus Apher Sci, 2018. 57(3): p. 422–424. pmid:29784537
- 16. Hamet P. and Tremblay J., Artificial intelligence in medicine. Metabolism, 2017. 69s: p. S36–s40. pmid:28126242
- 17. Hao H., et al., A paradigm for high-throughput screening of cell-selective surfaces coupling orthogonal gradients and machine learning-based cell recognition. Bioact Mater, 2023. 28: p. 1–11. pmid:37214260
- 18.
Wang L., Zhang H.-C., and Wang Q. On the concepts of artificial intelligence and innovative design in product design. in IOP conference series: materials science and engineering. 2019. IOP Publishing.
- 19.
Tavares J., et al., Cyber Intelligence and Information Retrieval. 2022: Springer.
- 20. Taye M.M., Understanding of Machine Learning with Deep Learning: Architectures, Workflow, Applications and Future Directions. Computers, 2023. 12(5): p. 91.
- 21.
Graves A. and Graves A., Supervised sequence labelling. 2012: Springer.
- 22.
Medsker L. and Jain L.C., Recurrent neural networks: design and applications. 1999: CRC press.
- 23. Choi R.Y., et al., Introduction to Machine Learning, Neural Networks, and Deep Learning. Transl Vis Sci Technol, 2020. 9(2): p. 14. pmid:32704420
- 24. Srinivasan M., et al., Exploring the Current Trends of Artificial Intelligence in Stem Cell Therapy: A Systematic Review. Cureus, 2021. 13(12): p. e20083.
- 25. Page M.J., et al., The PRISMA 2020 statement: an updated guideline for reporting systematic reviews. International journal of surgery, 2021. 88: p. 105906. pmid:33789826
- 26. Walsh I., et al., DOME: recommendations for supervised machine learning validation in biology. Nature Methods, 2021. 18(10): p. 1122–1127. pmid:34316068
- 27. Sun A., et al., 3D in vivo Magnetic Particle Imaging of Human Stem Cell-Derived Islet Organoid Transplantation Using a Machine Learning Algorithm. Front Cell Dev Biol, 2021. 9: p. 704483. pmid:34458264
- 28. Skorska A., et al., Monitoring the maturation of the sarcomere network: a super-resolution microscopy-based approach. Cell Mol Life Sci, 2022. 79(3): p. 149. pmid:35199227
- 29. Vuidel A., et al., High-content phenotyping of Parkinson’s disease patient stem cell-derived midbrain dopaminergic neurons using machine learning classification. Stem Cell Reports, 2022. 17(10): p. 2349–2364. pmid:36179692
- 30. Williams B., et al., Prediction of Human Induced Pluripotent Stem Cell Cardiac Differentiation Outcome by Multifactorial Process Modeling. Front Bioeng Biotechnol, 2020. 8: p. 851. pmid:32793579
- 31. Jiang B., et al. Application of Support Vector Machine to Recognize Trans-differentiated Neural Progenitor Cells for Bright-Field Microscopy. in 2015 Fifth International Conference on Instrumentation and Measurement, Computer, Communication and Control (IMCCC). 2015.
- 32.
Fischbacher B., et al., Modular deep learning enables automated identification of monoclonal cell lines. bioRxiv, 2020: p. 2020.12.28.424610.
- 33. Patino C.A., et al., Deep Learning and Computer Vision Strategies for Automated Gene Editing with a Single-Cell Electroporation Platform. SLAS Technol, 2021. 26(1): p. 26–36. pmid:33449846
- 34. Patino C.A., et al., Multiplexed high-throughput localized electroporation workflow with deep learning-based analysis for cell engineering. Sci Adv, 2022. 8(29): p. eabn7637. pmid:35867793
- 35. Hsu C.C., et al., A single-cell Raman-based platform to identify developmental stages of human pluripotent stem cell-derived neurons. Proc Natl Acad Sci U S A, 2020. 117(31): p. 18412–18423. pmid:32694205
- 36. Lien C.Y., et al., Recognizing the Differentiation Degree of Human Induced Pluripotent Stem Cell-Derived Retinal Pigment Epithelium Cells Using Machine Learning and Deep Learning-Based Approaches. Cells, 2023. 12(2). pmid:36672144
- 37. Verzat C., et al., Image-based deep learning reveals the responses of human motor neurons to stress and VCP-related ALS. Neuropathol Appl Neurobiol, 2022. 48(2): p. e12770. pmid:34595747
- 38. Kusumoto D., et al., Automated Deep Learning-Based System to Identify Endothelial Cells Derived from Induced Pluripotent Stem Cells. Stem Cell Reports, 2018. 10(6): p. 1687–1695. pmid:29754958
- 39. Joy D.A., Libby A.R.G., and McDevitt T.C., Deep neural net tracking of human pluripotent stem cells reveals intrinsic behaviors directing morphogenesis. Stem Cell Reports, 2021. 16(5): p. 1317–1330. pmid:33979602
- 40. Teles D., et al., Machine Learning Techniques to Classify Healthy and Diseased Cardiomyocytes by Contractility Profile. ACS Biomater Sci Eng, 2021. 7(7): p. 3043–3052. pmid:34152732
- 41. Guo J., et al., Machine learning-assisted high-content analysis of pluripotent stem cell-derived embryos in vitro. Stem Cell Reports, 2021. 16(5): p. 1331–1346.
- 42. Zhang H., et al., A novel machine learning based approach for iPS progenitor cell identification. PLoS Comput Biol, 2019. 15(12): p. e1007351. pmid:31877128
- 43. Joutsijoki H., et al. Histogram-based classification of iPSC colony images using machine learning methods. in 2014 IEEE International Conference on Systems, Man, and Cybernetics (SMC). 2014.
- 44. Joutsijoki H., et al., Machine Learning Approach to Automated Quality Identification of Human Induced Pluripotent Stem Cell Colony Images. Comput Math Methods Med, 2016. 2016: p. 3091039. pmid:27493680
- 45. Molugu K., et al., Label-Free Imaging to Track Reprogramming of Human Somatic Cells. GEN Biotechnol, 2022. 1(2): p. 176–191. pmid:35586336
- 46. Tokunaga K., et al., Computational image analysis of colony and nuclear morphology to evaluate human induced pluripotent stem cells. Sci Rep, 2014. 4: p. 6996. pmid:25385348
- 47. Fan K., et al., A Machine Learning Assisted, Label-free, Non-invasive Approach for Somatic Reprogramming in Induced Pluripotent Stem Cell Colony Formation Detection and Prediction. Scientific Reports, 2017. 7(1): p. 13496.
- 48. Ye K., et al., Reproducible production and image-based quality evaluation of retinal pigment epithelium sheets from human induced pluripotent stem cells. Scientific Reports, 2020. 10(1): p. 14387. pmid:32873827
- 49. Orita K., et al., Machine-learning-based quality control of contractility of cultured human-induced pluripotent stem-cell-derived cardiomyocytes. Biochem Biophys Res Commun, 2020. 526(3): p. 751–755. pmid:32265031
- 50. Orita K., et al., Deep learning-based quality control of cultured human-induced pluripotent stem cell-derived cardiomyocytes. J Pharmacol Sci, 2019. 140(4): p. 313–316. pmid:31113731
- 51. Park K., et al., Deep learning predicts the differentiation of kidney organoids derived from human induced pluripotent stem cells. Kidney Res Clin Pract, 2023. 42(1): p. 75–85. pmid:36328994
- 52. Powell K.A., et al., Automated human induced pluripotent stem cell colony segmentation for use in cell culture automation applications. SLAS Technol, 2023. 28(6): p. 416–422. pmid:37454765
- 53. Yang L., et al., High-Content Screening and Analysis of Stem Cell-Derived Neural Interfaces Using a Combinatorial Nanotechnology and Machine Learning Approach. Research (Wash D C), 2022. 2022: p. 9784273. pmid:36204248
- 54. Maddah M. and Loewke K., Automated, non-invasive characterization of stem cell-derived cardiomyocytes from phase-contrast microscopy. Med Image Comput Comput Assist Interv, 2014. 17(Pt 1): p. 57–64. pmid:25333101
- 55. Kim M., et al., Prediction of Stem Cell State Using Cell Image-Based Deep Learning. Advanced Intelligent Systems, 2023. 5(7): p. 2300017.
- 56. Kavitha M.S., Kurita T., and Ahn B.C., Critical texture pattern feature assessment for characterizing colonies of induced pluripotent stem cells through machine learning techniques. Comput Biol Med, 2018. 94: p. 55–64. pmid:29407998
- 57. Ta N., et al., Mining Key Regulators of Cell Reprogramming and Prediction Research Based on Deep Learning Neural Networks. IEEE Access, 2020. 8: p. 23179–23185.
- 58. Mukherjee P., et al., Deep Learning-Assisted Automated Single Cell Electroporation Platform for Effective Genetic Manipulation of Hard-to-Transfect Cells. Small, 2022. 18(20): p. e2107795. pmid:35315229
- 59. Smith Q., et al., Cytoskeletal tension regulates mesodermal spatial organization and subsequent vascular fate. Proceedings of the National Academy of Sciences, 2018. 115(32): p. 8167–8172. pmid:30038020
- 60. Esmail S. and Danter W.R., DeepNEU: Artificially Induced Stem Cell (aiPSC) and Differentiated Skeletal Muscle Cell (aiSkMC) Simulations of Infantile Onset POMPE Disease (IOPD) for Potential Biomarker Identification and Drug Discovery. Front Cell Dev Biol, 2019. 7: p. 325. pmid:31867331
- 61. Esmail S. and Danter W., Viral pandemic preparedness: A pluripotent stem cell-based machine-learning platform for simulating SARS-CoV-2 infection to enable drug discovery and repurposing. Stem Cells Transl Med, 2021. 10(2): p. 239–250. pmid:32961040
- 62. Aida S., et al., Conditional Generative Adversarial Networks to Model iPSC-Derived Cancer Stem Cells. Journal of Advanced Computational Intelligence and Intelligent Informatics, 2020. 24(1): p. 134–141.
- 63. Atwell S., et al., Label-free imaging of 3D pluripotent stem cell differentiation dynamics on chip. Cell Rep Methods, 2023. 3(7): p. 100523. pmid:37533640
- 64. Chu S.L., et al., Human induced pluripotent stem cell formation and morphology prediction during reprogramming with time-lapse bright-field microscopy images using deep learning methods. Comput Methods Programs Biomed, 2023. 229: p. 107264. pmid:36473419
- 65. Wakui T., et al., Method for evaluation of human induced pluripotent stem cell quality using image analysis based on the biological morphology of cells. J Med Imaging (Bellingham), 2017. 4(4): p. 044003. pmid:29134187
- 66. Piotrowski T., et al., Deep-learning-based multi-class segmentation for automated, non-invasive routine assessment of human pluripotent stem cell culture status. Comput Biol Med, 2021. 129: p. 104172. pmid:33352307
- 67. Iwagawa T., et al., Evaluation of CRISPR/Cas9 exon-skipping vector for choroideremia using human induced pluripotent stem cell-derived RPE. J Gene Med, 2023. 25(2): p. e3464. pmid:36413603
- 68. Hayashi Y., et al., Automated adherent cell elimination by a high-speed laser mediated by a light-responsive polymer. Communications Biology, 2018. 1(1): p. 218. pmid:30534610
- 69. Yuan-Hsiang C., et al., Human induced pluripotent stem cell region recognition in microscopy images using Convolutional Neural Networks. Annu Int Conf IEEE Eng Med Biol Soc, 2017. 2017: p. 4058–4061. pmid:29060788
- 70. Danter W.R., DeepNEU: cellular reprogramming comes of age—a machine learning platform with application to rare diseases research. Orphanet J Rare Dis, 2019. 14(1): p. 13. pmid:30630505
- 71. Catanese A., et al., Multiomics and machine-learning identify novel transcriptional and mutational signatures in amyotrophic lateral sclerosis. Brain, 2023. 146(9): p. 3770–3782. pmid:36883643
- 72. Sekiya A., et al., Variation of DNA methylation on the IRX1/2 genes is responsible for the neural differentiation propensity in human induced pluripotent stem cells. Regen Ther, 2022. 21: p. 620–630. pmid:36514370
- 73. Lai B., et al., Annotating functional effects of non-coding variants in neuropsychiatric cell types by deep transfer learning. PLoS Comput Biol, 2022. 18(5): p. e1010011. pmid:35576194
- 74. Bardy C., et al., Predicting the functional states of human iPSC-derived neurons with single-cell RNA-seq and electrophysiology. Mol Psychiatry, 2016. 21(11): p. 1573–1588. pmid:27698428
- 75. Theodoris C.V., et al., Network-based screen in iPSC-derived cells reveals therapeutic candidate for heart valve disease. Science, 2021. 371(6530). pmid:33303684
- 76. Nishino K., et al., Identification of an epigenetic signature in human induced pluripotent stem cells using a linear machine learning model. Hum Cell, 2021. 34(1): p. 99–110. pmid:33047283
- 77. Nguyen Q.H., et al., Single-cell RNA-seq of human induced pluripotent stem cells reveals cellular heterogeneity and cell state transitions between subpopulations. Genome Res, 2018. 28(7): p. 1053–1066. pmid:29752298
- 78. Liu S.J., et al., CRISPRi-based genome-scale identification of functional long noncoding RNA loci in human cells. Science, 2017. 355(6320). pmid:27980086
- 79. Nguyen T.B., et al., Harshening stem cell research and precision medicine: The states of human pluripotent cells stem cell repository diversity, and racial and sex differences in transcriptomes. Front Cell Dev Biol, 2022. 10: p. 1071243. pmid:36684445
- 80. Feng W., et al., Computational profiling of hiPSC-derived heart organoids reveals chamber defects associated with NKX2-5 deficiency. Communications Biology, 2022. 5(1): p. 399. pmid:35488063
- 81. Joutsijoki H., et al., Separation of HCM and LQT Cardiac Diseases with Machine Learning of Ca2+ Transient Profiles. Methods Inf Med, 2019. 58(4–05): p. 167–178. pmid:32079026
- 82. Yang H., et al., Deriving waveform parameters from calcium transients in human iPSC-derived cardiomyocytes to predict cardiac activity with machine learning. Stem Cell Reports, 2022. 17(3): p. 556–568. pmid:35148844
- 83. Hwang H., et al., Machine learning identifies abnormal Ca(2+) transients in human induced pluripotent stem cell-derived cardiomyocytes. Sci Rep, 2020. 10(1): p. 16977. pmid:33046816
- 84. Pang J.K.S., et al., Characterizing arrhythmia using machine learning analysis of Ca(2+) cycling in human cardiomyocytes. Stem Cell Reports, 2022. 17(8): p. 1810–1823.
- 85. Juhola M., et al., Detection of genetic cardiac diseases by Ca2+ transient profiles using machine learning methods. Scientific Reports, 2018. 8(1): p. 9355. pmid:29921843
- 86. Schaub N.J., et al., Deep learning predicts function of live retinal pigment epithelium from quantitative microscopy. J Clin Invest, 2020. 130(2): p. 1010–1023. pmid:31714897
- 87. Aghasafari P., et al., A deep learning algorithm to translate and classify cardiac electrophysiology. Elife, 2021. 10. pmid:34212860
- 88. Tripathi U., et al., Information theory characteristics improve the prediction of lithium response in bipolar disorder patients using a support vector machine classifier. Bipolar Disord, 2023. 25(2): p. 110–127. pmid:36479788
- 89. Witmer A. and Bhanu B., Multi-label Classification of Stem Cell Microscopy Images Using Deep Learning. 2018. 1408–1413.
- 90. Imamura K., et al., Prediction Model of Amyotrophic Lateral Sclerosis by Deep Learning with Patient Induced Pluripotent Stem Cells. Ann Neurol, 2021. 89(6): p. 1226–1233. pmid:33565152
- 91. Juhola M., et al., On computational classification of genetic cardiac diseases applying iPSC cardiomyocytes. Comput Methods Programs Biomed, 2021. 210: p. 106367. pmid:34474196
- 92. Juhola M., et al., A method to measure data complexity of a complicated medical data set. International Journal of Imaging Systems and Technology, 2022. 32(6): p. 1822–1831.
- 93. Kowalczewski A., et al., Integrating nonlinear analysis and machine learning for human induced pluripotent stem cell-based drug cardiotoxicity testing. J Tissue Eng Regen Med, 2022. 16(8): p. 732–743. pmid:35621199
- 94. Monzel A.S., et al., Machine learning-assisted neurotoxicity prediction in human midbrain organoids. Parkinsonism Relat Disord, 2020. 75: p. 105–109. pmid:32534431
- 95. Heylman C., et al., Supervised Machine Learning for Classification of the Electrophysiological Effects of Chronotropic Drugs on Human Induced Pluripotent Stem Cell-Derived Cardiomyocytes. PLoS One, 2015. 10(12): p. e0144572. pmid:26695765
- 96. Lee E.K., et al., Machine learning plus optical flow: a simple and sensitive method to detect cardioactive drugs. Sci Rep, 2015. 5: p. 11817. pmid:26139150
- 97. Grafton F., et al., Deep learning detects cardiotoxicity in a high-content screen with induced pluripotent stem cell-derived cardiomyocytes. Elife, 2021. 10.
- 98. Yang H., et al., Prediction of inotropic effect based on calcium transients in human iPSC-derived cardiomyocytes and machine learning. Toxicol Appl Pharmacol, 2023. 459: p. 116342. pmid:36502871
- 99. Kandasamy K., et al., Prediction of drug-induced nephrotoxicity and injury mechanisms with human induced pluripotent stem cell-derived cells and machine learning methods. Sci Rep, 2015. 5: p. 12337. pmid:26212763
- 100. Jaklin M., et al., Optimization of the TeraTox Assay for Preclinical Teratogenicity Assessment. Toxicol Sci, 2022. 188(1): p. 17–33. pmid:35485993
- 101. Maddah M., et al., Quantifying drug-induced structural toxicity in hepatocytes and cardiomyocytes derived from hiPSCs using a deep learning method. J Pharmacol Toxicol Methods, 2020. 105: p. 106895. pmid:32629158
- 102. Juhola M., et al., Analysis of Drug Effects on iPSC Cardiomyocytes with Machine Learning. Ann Biomed Eng, 2021. 49(1): p. 129–138. pmid:32367466
- 103. Matsuda N., et al., Raster plots machine learning to predict the seizure liability of drugs and to identify drugs. Sci Rep, 2022. 12(1): p. 2281. pmid:35145132
- 104. Hidaka T., et al., Prediction of Compound Bioactivities Using Heat-Diffusion Equation. Patterns (N Y), 2020. 1(9): p. 100140. pmid:33336198
- 105. Hanafusa Y., Shiraishi A., and Hattori F., Machine learning discriminates P2X7-mediated intracellular calcium sparks in human-induced pluripotent stem cell-derived neural stem cells. Scientific Reports, 2023. 13(1): p. 12673. pmid:37542080
- 106. Rao M.S. and Malik N., Assessing iPSC reprogramming methods for their suitability in translational medicine. J Cell Biochem, 2012. 113(10): p. 3061–8. pmid:22573568
- 107. Strano A., et al., Variable Outcomes in Neural Differentiation of Human PSCs Arise from Intrinsic Differences in Developmental Signaling Pathways. Cell Rep, 2020. 31(10): p. 107732. pmid:32521257
- 108. Koyanagi-Aoi M., et al., Differentiation-defective phenotypes revealed by large-scale analyses of human pluripotent stem cells. Proc Natl Acad Sci U S A, 2013. 110(51): p. 20569–74. pmid:24259714
- 109. Nosrati H. and Nosrati M., Artificial Intelligence in Regenerative Medicine: Applications and Implications. Biomimetics (Basel), 2023. 8(5).
- 110. Martins-Taylor K. and Xu R.H., Concise review: Genomic stability of human induced pluripotent stem cells. Stem Cells, 2012. 30(1): p. 22–7. pmid:21823210
- 111. Volpato V. and Webber C., Addressing variability in iPSC-derived models of human disease: guidelines to promote reproducibility. Dis Model Mech, 2020. 13(1). pmid:31953356
- 112. Lee B., et al., Cell Culture Process Scale-Up Challenges for Commercial-Scale Manufacturing of Allogeneic Pluripotent Stem Cell Products. Bioengineering (Basel), 2022. 9(3). pmid:35324781
- 113. Tristan C.A., et al., Robotic high-throughput biomanufacturing and functional differentiation of human pluripotent stem cells. Stem Cell Reports, 2021. 16(12): p. 3076–3092.
- 114. Bardy C., et al., Predicting the functional states of human iPSC-derived neurons with single-cell RNA-seq and electrophysiology. Molecular Psychiatry, 2016. 21(11): p. 1573–1588. pmid:27698428
- 115. Kusumoto D., Yuasa S., and Fukuda K., Induced Pluripotent Stem Cell-Based Drug Screening by Use of Artificial Intelligence. Pharmaceuticals, 2022. 15(5): p. 562. pmid:35631387
- 116. Hanafusa Y., Shiraishi A., and Hattori F., Machine learning discriminates P2X7-mediated intracellular calcium sparks in human-induced pluripotent stem cell-derived neural stem cells. Sci Rep, 2023. 13(1): p. 12673. pmid:37542080
- 117. Kusumoto D., et al., Anti-senescent drug screening by deep learning-based morphology senescence scoring. Nat Commun, 2021. 12(1): p. 257. pmid:33431893
- 118. Wang X., et al., SSGraphCPI: A novel model for predicting compound-protein interactions based on deep learning. International Journal of Molecular Sciences, 2022. 23(7): p. 3780. pmid:35409140
- 119. Maddah M., et al., Quantifying drug-induced structural toxicity in hepatocytes and cardiomyocytes derived from hiPSCs using a deep learning method. Journal of Pharmacological and Toxicological Methods, 2020. 105: p. 106895. pmid:32629158
- 120.
Gilpin L.H., et al. Explaining explanations: An overview of interpretability of machine learning. in 2018 IEEE 5th International Conference on data science and advanced analytics (DSAA). 2018. IEEE.
- 121. Laine R.F., et al., Avoiding a replication crisis in deep-learning-based bioimage analysis. Nat Methods, 2021. 18(10): p. 1136–1144. pmid:34608322
- 122. Jones D.T., Setting the standards for machine learning in biology. Nat Rev Mol Cell Biol, 2019. 20(11): p. 659–660. pmid:31548714
- 123. Barberis A., Aerts H., and Buffa F.M., Robustness and reproducibility for AI learning in biomedical sciences: RENOIR. Sci Rep, 2024. 14(1): p. 1933. pmid:38253545