Introduction
The incorporation and implementation of Artificial Intelligence (AI) into Umbilical cord biobanking is still at its developing stage, but there is a great potential that artificial intelligence will be progressively incorporated and implemented into the various aspects of umbilical cord blood biobanking. This article elaborates extensively on the various aspects of the Umbilical Cord biobanking where Artificial intelligence could play pivotal roles.
Schematic outlining the AI areas
Figure 1: Schematic outlining the AI areas (Figure courtesy De Antonio M et al. 2020)
Key Areas in the Umbilical Cord Biobanking Where Artificial Intelligence Can Be Implemented
Informed Consent
AI-based systems which include machine-learning and natural language processing methods can be designed and implemented to understand and explain the contents of consent forms and manage web-based communications with umbilical cord biobank participants. If a participant chooses to withdraw, the AI system could destroy any associated data and notify the biobank’s administrators to dispose of the corresponding biological specimen.
Sample Management
The incorporation of an AI system into an automated sample storage system, can assist in repositioning bio-samples to utilize vacant spaces within the storage system more efficiently. AI may also be designed and used to develop standard operating protocols (SOPs) tailored to specific biological sample uses and to identify or match a biospecimen appropriately to a specific study. AI can also initiate a biospecimen collection plan for prospective biomedical research depending on its analysis of the biobank’s distribution and inventory status, as well as research trends.
Featured Partners
Human Leukocyte Antigen Typing (HLA)
HLA typing is a requirement by regulatory bodies for Umbilical cord biobanks. Deep learning and machine learning techniques have been created to estimate HLA genotypes and matching patients with suitable cord blood units for transplantation. This process is known to be faster and more cost-effective than laboratory based HLA genotyping methods such as direct sequencing, allele-specific amplification, or hybridization, making it particularly beneficial for large-scale datasets.
The HLA*IMP framework which was developed by Dilthey et al. 2011 utilizes genotypic data from various genome-wide single nucleotide polymorphism (SNP) datasets to impute classical HLA alleles with an accuracy of 92–98%, comparable to lab-based HLA-typing techniques. Furthermore, HLA*IMP:02 which is an advancement to the HLA*IMP was designed to use Single nucleotide polymorphism (SNP) data from diverse population and ethnicity settings to account for genotypic heterogeneity, achieving an imputation accuracy of 90–97% across European and non-European panels.
The HLA Genotype Imputation and Attribute Bagging (HIBAG) software program was later developed, and has an edge over HLA*IMP in accuracy performance. Additionally, DEEP*HLA, a convolutional neural network (CNN), was created to estimate low-frequency and rare HLA-alleles with high accuracy and the shortest processing time, making it ideal for biobank-scale data.
Diagnostic Screening
Diagnostic screening of infectious diseases from maternal blood samples is a required safety test for Umbilical cord samples. AI can assist in this area. A study recently carried out by Mohammad M et al. 2022, explored the use of a stacked ensemble, which combines traditional parametric (logistic regression) and non-parametric machine learning methods (random forest and gradient boosting trees), to effectively screen for the hepatitis C virus (HCV), which is often under-diagnosed. This method achieved a precision rate of 97%, which is 66% more precise than conventional logistic regression.
AI can also be used in cytogenetics for karyotyping analysis to further ensure sample safety. Ikaros, a karyotyping software based on deep neural networks, can accurately analyze chromosomes, and reduce processing times.
A novel AI-based method known as ChromoEnhancer, adopts CycleGAN (Generative Adversarial Networks) to enhance images of neoplastic karyograms, thereby assisting in the accurate detection of hidden chromosomal abnormalities.
Provision of Genomic Data
In umbilical cord biobank that serves as a genomic data provider, the use of “sure independence screening” (SIS), a statistical machine learning method, could be of great importance. SIS can be used in Single nucleotide polymorphism (SNP) analysis for genome-wide association studies (GWAS) and in analyzing genome-wide gene-gene interactions.
In a study carried out by Narita et al. 2021, the SIS algorithm was incorporated into a software program known as EPISIS which was able to identify functional and epistatic interactions between genetic components linked to a severe form of Stevens-Johnson Syndrome. EPISIS has the capability of conducting significant gene-gene interaction analysis in biobanks.
Other machine learning methods which can effectively predict phenotypes and identify genetic risk factors for specific diseases or conditions include neural networks, lasso regression, support vector machines (SVM), and random forest. ExPecto which is a recent development is a scalable deep learning approach which can accurately predict common or rare variants within disease or trait-associated loci based on sequence data.
Error Detection in Data Associated with a Biological Sample
There are innovative machine-learning techniques designed to identify outliers, or potential errors, in data or questionnaires linked to a biological sample. kurPCA and RAMP were designed for this purpose. kurPCA merges Principal Component Analysis (PCA) and kurtosis to identify these errors while RAMP (Regression Adjustment Method for Systematic Missing Pattern) detects errors and simplifies the data cleaning process, which would have required a great amount of manual effort. These techniques help to improve the precision and efficiency of data analysis.
Quality Assessment
AI can be used to aid quality assessment of Umbilical cord mesenchymal stem cells (UC-MSCs) and also for streamlining the biobanking and cell therapy process.
In a study by Marklein et al, 2019. A machine learning approach called visual stochastic neighbor embedding (viSNE) was used to accurately identify gamma interferon (IFN-γ-) stimulated morphological subpopulations of MSCs. These subpopulations, which were unidentifiable using single or multiple morphological analysis, were identified at a single-cell resolution, and strongly corresponded with their immunosuppressive capacity. This can help better identify the functionality of MSCs population, and consequently their therapeutic potential.
In another study by Mota et al. 2021 a machine learning algorithm was designed to determine the viability of MSCs according to their morphological phenotype and characterize them as either self-replicating or slowly replicating cells. Slowly replicating cells had impaired differentiation capabilities, and therefore impaired therapeutic potential. Based on their findings AI may be incorporated to identify MSCs based on their differentiation potential thereby providing a non-invasive cell quality assessment.
Predictive Modeling
Data mining approaches such as decision trees, artificial neural networks (ANN) and SVM, can be potentially used to develop a prediction model based on attribute selection (diagnosis, age, gender, and conditioning protocol during transplantation) for treatment-related morbidity and mortality post-allogeneic Human stem cell transplant (HSCT) thereby aiding in donor and patient selection for treatment and improving treatment outcome.
A study by Giovanni et al. observed ANNs sensitivity in predicting acute graft-vs-host disease (GvHD) post HSC transplantation to be 83.3%, and 90.1% in predicting the absence of patients developing acute GvHD.
Conclusion
Artificial intelligence (AI) is yet to be widely adopted and implemented in Umbilical cord biobanking, and as a result of this, its definite potential remains to be fully utilized. The implementation of AI in cord biobanking has the potential to revolutionize the Umbilical cord biobanking, enabling more efficient and effective use of cord blood stem cells for transplantation and research. Investigating and further validating appropriate Artificial intelligence algorithms relating to Umbilical cord biobanking more specifically, is essential to expand the applicability of AI into the field.
References
Annaratone L, De Palma G, Bonizzi G, et al. 2021. Basic principles of biobanking: from biological samples to precision medicine for patients. Virchows Archiv. 2021;479(2):233–246.
Ballen KK, Verter F, Kurtzberg J. 2012. Umbilical cord blood donation: public or private? Bone Marrow Transplant. 2015;50:1271–1278.
Battineni G, Hossain MA, Chintalapudi N, Amenta F. 2022. A survey on the role of artificial intelligence in biobanking studies: a systematic review. Diagnostics. 2022;12:1179.
Bokhari Y, Alhareeri A, Aljouie A, et al.2022.ChromoEnhancer: an artificial-intelligence-based tool to enhance neoplastic karyograms as an aid for effective analysis. Cells. 2022;11:2244.
Butler MG, Menitove JE.2011. Umbilical cord blood banking: an update. J Assist Reprod Genet. 2011;28:669–676.
Caocci G, Baccoli R, Vacca A, et al.2010. Comparison between an artificial neural network and logistic regression in predicting acute graft-vs-host disease after unrelated donor hematopoietic stem cell transplantation in thalassemia patients. Exp Hematol. 2010;38:426–433.
Cirillo D, Valencia A.2019. Big data analytics for personalized medicine. Curr Opin Biotechnol. 2019;58:161–167.
Committee on Establishing a National Cord Blood Stem Cell Bank Program, et al. 2005. Cord Blood: Establishing a National Hematopoietic Stem Cell Bank Program. National Academies Press; 2005.
Dagher G.2022. Quality matters: international standards for biobanking. Cell Prolif. 2022;55.
De Antonio M., Timothy J.W.D., James Philip H., O’Regan D.P 2020. Artificial intelligence and the cardiologist: What you need to know for 2020. Heart. 2020;106:39
Dilthey AT, Moutsianas L, Leslie S, McVean G.2011. HLA*IMP—an integrated framework for imputing classical HLA alleles from SNP genotypes. Bioinformatics. 2011;27:968–972.
Dilthey A, Leslie S, Moutsianas L, et al.2013. Multi-population classical HLA type imputation.PLoS Comput Biol. 2013;9:e1002877.
Knoppers BM, Isasi R. 2010. Stem cell banking: between traceability and identifiability. Genome Med. 2010;2:73.
Kim G, Jeon JH, Park K, et al.2022. High throughput screening of mesenchymal stem cell lines using deep learning. Sci Rep. 2022;12:17507.
Lee J-E. 2018. Artificial intelligence in the future biobanking: current issues in the biobank and future possibilities of artificial intelligence. Biomed J Sci Tech Res. 2018;7:5937–9.
Marklein RA, Klinker MW, Drake KA, et al.2019. Morphological profiling using machine learning reveals emergent subpopulations of interferon-γ–stimulated mesenchymal stromal cells that predict immunosuppression. Cytotherapy. 2019;21:17–31.
Matsuoka F, Takeuchi I, Agata H, et al.2013. Morphology-based prediction of osteogenic differentiation potential of human mesenchymal stem cells. PLoS One. 2013;8:e55082.
Mahajan A, Vaidya T, Gupta A, Rane S, Gupta S.2019. Artificial intelligence in healthcare in developing nations: the beginning of a transformative journey. Cancer Res Stat Treat. 2019;2:182.
Mota SM, Rogers RE, Haskell AW, et al.2021. Automated mesenchymal stem cell segmentation and machine learning-based phenotype classification using morphometric and textural analysis. J Med Imaging. 2021;8.
Mohammad MM, Sepideh N, Roya NV, Ali AH, Hossein M. 2022. Prediction of Hepatitis disease using ensemble learning methods.J Prev Hyg. 2022: 63 (3): E424-E428.
Naito T, Suzuki K, Hirata J, et al. 2021. A deep learning method for HLA imputation and trans-ethnic MHC fine-mapping of type 1 diabetes. Nat Commun. 2021;12:1639.
Narita A, Ueki M, Tamiya G. 2021. Artificial intelligence powered statistical genetics in biobanks. J Hum Genet. 2021;66:61–65.
Sakurai R, Ueki M, Makino S, et al. 2019. Outlier detection for questionnaire data in biobanks. Int J Epidemiol. 2019;48:1305–1315.
Shouval R, Bondi O, Mishan H, et al. 2014. Application of machine learning algorithms for clinical predictive modeling: a data-mining approach in SCT. Bone Marrow Transplant. 2014;49:332–337.
Yan Q, Jiang Y, Huang H, et al. 2021. Genome-wide association studies-based machine learning for prediction of age-related macular degeneration risk. Transl Vis Sci Technol. 2021;10:29.
Zhou J, Theesfeld CL, Yao K, et al.2018. Deep learning sequence-based ab initio prediction of variant effects on expression and disease risk. Nat Genet. 2018;50:1171–1179.
About the Author
Tijani Abdulrahman Debo
+ Posts
Bsc. Zoology, BMLS, Msc. Biobanks & Complex data management (In view)
Tijani Abdulrahman is a medical laboratory scientist and a biobanking professional with 10+ years of demonstrated history in working in the medical science practice industry. Skilled in laboratory medicine, biobanking, quality management system and biomedical research. Currently a master’s student at the University Cote d’Azur, France with focus on Biobanking and complex data management. With a multidisciplinary background Tijani Abdulrahman strives to champion biobank and biomedical enabled research in order to strengthen biobank and biomedical innovation.
Stay Ahead in Biobanking
Get valuable biobanking news, insights and resources delivered to your inbox for free!
Your work email
Subscribe →
No spam, unsubscribe anytime.
TAGSAIArtificial IntelligenceCord Blood BankingStem cellsUmbilical CordUmbilical Cord Biobanking
Share
RELATED ARTICLESMORE FROM AUTHOR
Brain-Derived p-Tau and Biobanks: A Q&A with Prof. Henrik Zetterberg on Next-Generation Alzheimer’s Blood Biomarkers
Insights & Opinions
Brain-Derived p-Tau and Biobanks: A Q&A with Prof. Henrik Zetterberg on Next-Generation Alzheimer’s Blood Biomarkers
Room Temperature Biobanking Can you Imagine Preserving Your Samples with Maximum Quality Without Freezing?
Blood & Biofluids Banking
Room Temperature Biobanking: Can you Imagine Preserving Your Samples with Maximum Quality Without Freezing?
Democratizing Biobanking: Making Cutting-Edge Research Accessible to Emerging Countries
Industry Insights
Democratizing Biobanking: Making Cutting-Edge Research Accessible to Emerging Countries
