Advancements in artificial intelligence and machine learning in revolutionising biomarker discovery

Raikar, Gokuldas (Vedant) Sarvesh; Raikar, Amisha Sarvesh; Somnache, Sandesh Narayan

doi:10.1590/s2175-97902023e23146

Acessibilidade / Reportar erro

Brasil

Brazilian Journal of Pharmaceutical Sciences

Español English

Brasil

Español English

sumário « anterior atual seguinte »

Sumário

Article • Braz. J. Pharm. Sci. 59 • 2023 • https://doi.org/10.1590/s2175-97902023e23146 copy

Advancements in artificial intelligence and machine learning in revolutionising biomarker discovery

Authorship SCIMAGO INSTITUTIONS RANKINGS

Abstract

The article explores the significance of biomarkers in clinical research and the advantages of utilizing artificial intelligence (AI) and machine learning (ML) in the discovery process. Biomarkers provide a more comprehensive understanding of disease progression and response to therapy compared to traditional indicators. AI and ML offer a new approach to biomarker discovery, leveraging large amounts of data to identify patterns and optimize existing biomarkers. Additionally, the article touches on the emergence of digital biomarkers, which use technology to assess an individual’s physiological and behavioural states, and the importance of properly processing omics and multi-omics data for efficient handling by computer systems. However, the article acknowledges the challenges posed by AI/ML in the identification of biomarkers, including potential biases in the data and the need for diversity in data representation. To address these challenges, the article suggests the importance of regulation and diversity in the development of AI/ML algorithms.

Keywords:
Biomarkers; Artificial Intelligence; Machine Learning; Multi-omics; Omics

Categories	Application	Example	Ref. No.
Diagnostic Digital Biomarker	To diagnose Medical Condition	Use of accelerometers and gyroscopes in smartphones and wearable devices for diagnosing and monitoring of neurological conditions such as Parkinson’s disease.	(Abou et al., 2021Abou L, Peters J, Wong E, Akers R, Dossou MS, Sosnoff JJ, et al. Gait and Balance Assessments using Smartphone Applications in Parkinson’s Disease: A Systematic Review. J Med Syst. 2021;45(9):87.)
Response Digital Biomarker	To evaluate impact of treatment	Use of smartphone-based digital biomarkers to monitor the response of patients to depression treatment. In this machine learning algorithms is used for analysis of data related to sign-symptoms, side effects, and adherence to treatment and predicting the response to the treatment.	(Mandryk et al., 2021Mandryk RL, Birk MV, Vedress S, Wiley K, Reid E, Berger P, et al. Remote Assessment of Depression Using Digital Biomarkers from Cognitive Tasks. Front Psychol. 2021;12:767507.)
Monitoring Digital Biomarker	To track changes in a person’s health over time	Use of wearable heart rate monitors for continuous monitoring of heart function.	(Singhal, Cowie, 2020Singhal A, Cowie MR. The role of wearables in heart failure. Curr Heart Failure Rep. 2020 Aug;17(4):125-132.)

Biomarkers	Biosensing mechanism	AI algorithms	Ref. No.
Physiological Signal	Wearable Device	Machine Learning	(*De Fazio et al., 2022*De Fazio R, Mattei V, Al-Naami B, De Vittorio M, Visconti P. Methodologies and Wearable Devices to Monitor Biophysical Parameters Related to Sleep Dysfunctions: An Overview. Micromachines. 2022;13(10):1335.)
Glucose Levels	CGM Device	Machine Learning for Real-Time Prediction	(*Galderisi et al., 2019*Galderisi A, Zammataro L, Losiouk E, Lanzola G, Kraemer K, Facchinetti A, et al. Continuous Glucose Monitoring Linked to an Artificial Intelligence Risk Index: Early Footprints of Intraventricular Hemorrhage in Preterm Neonates. Diabetes Technol Ther. 2019;21(3):146-153.)
ECG Signal	ECG Electrodes	Machine Learning for ECG Signal	(*Attia et al., 2021*Attia ZI, Harmon DM, Behr ER, Friedman PA. Application of artificial intelligence to the electrocardiogram. Eur Heart J. 2021;42(46):4717-4730.)
Blood-Based Signal	Blood Testing	Machine Learning	(*Putcha et al., 2020*Putcha G, Liu TY, Ariazi E, Bertin M, Drake A, Dzamba M, et al. Blood-based detection of early-stage colorectal cancer using multiomics and machine learning. In Abstract presented at: the American Society of Clinical Oncology (ASCO) GI Symposium 2020:23-25.)
Neuroimaging	Magnetic Resonance Imaging (MRI)	Machine Learning for Neuroimaging Analysis	(Razavi, Tarokh, Alborzi, 2019Razavi F, Tarokh MJ, Alborzi M. An intelligent Alzheimer’s disease diagnosis method using unsupervised feature learning. J Big Data. 2019;6:32.)

Type of Study	Biomarkers	Biosensing Mechanism	AI Algorithms	Ref. No.
Non-invasive measurement of glucose levels	Blood glucose	Reflectance spectroscopy levels	Artificial Neural Networks	(*Zhang et al., 2019*Zhang R, Liu S, Jin H, Luo Y, Zheng Z, Gao F, et al. Noninvasive Electromagnetic Wave Sensing of Glucose. Sensors. 2019;19(5):1151.)
Heart rate monitoring	Heart rate	Photoplethysmography	Convolutional Neural Networks	(*Coppetti et al., 2017*Coppetti T, Brauchlin A, Müggler S, Attinger TA, Templin C, Schönrath F, et al. Accuracy of smartphone apps for heart rate measurement. Eur J Prev Cardiol. 2017;24(12):1287-1293.)
Blood pressure monitoring	Blood pressure	Photoplethysmography	Deep Belief Networks	(*Luo et al., 2019*Luo H, Yang D, Barszczyk A, Vempala N, Wei J, Wu SJ, et al. Smartphone-Based Blood Pressure Measurement Using Transdermal Optical Imaging Technology. Circ Cardiovasc Imaging. 2019;12(8):e008857.)
Stress level monitoring	Stress level	Photoplethysmography, Electrodermal activity	Support Vector Machines	(Gupta, Alam, Agarwal, 2020Gupta R, Alam MA, Agarwal P. Modified Support Vector Machine for Detecting Stress Level Using EEG Signals. In: Versaci M, editor. Computational Intelligence and Neuroscience. 2020:8860841.)
Sleep quality monitoring	Sleep quality	Accelerometery, Gyroscope, Magnetometer	Decision Trees	(*Zhang et al., 2021a*Zhang C, Zeng P, Tan J, Sun S, Zhao M, Cui J, et al. Relationship of problematic smartphone use, sleep quality, and daytime fatigue among quarantined medical students during the COVID-19 pandemic. Front Psychiatry. 2021a;12:755059.)

Study type	Disease type	Tools used	Method language	Input data used	No. of samples	Input data source	Ref.
Multiomics analysis for Alzheimer’s disease biomarker prediction	Alzheimer’s disease	MOFA (Multi-Omics Factor Analysis)	MOFA package in R and Python software.	metabolomics and proteomics data	120	Alzheimer’s Research and therapy	(*Clark et al., 2021*Clark C, Dayon L, Masoodi M, Bowman GL, Popp J. An integrative multi-omics approach reveals new central nervous system pathway alterations in Alzheimer’s disease. Alzheimers Res Ther. 2021;13(1):71.)
Integrative Multiomics analysis for prostate cancer biomarker prediction	Prostate cancer	Interrogative Biology platform, Bayesian Network Inference (BNI) modules, Elastic-net coupled with bootstrapping.	R	Proteomics, metabolomics, and phospholipids	382	Journal of translation medicine	(*Kiebish et al., 2020*Kiebish AM, Cullen J, Mishra P, Ali A, Milliman E, Rodrigues LO, et al. multi-omic serum biomarkers for prognosis of disease progression in prostate cancer. J Transl Med. 2020;18:10.)
Multiomics analysis for colorectal cancer biomarker prediction	Colorectal cancer	Enrichment analyses of the co-expressed proteins	Python	proteomic and transcriptomic datasets	538	Journal of translation medicine	(*Zhang et al., 2021a*Zhang W, Lin L, Xia L, Cai W, Dai W, Zou C, et al. Multi-omics analyses of human colorectal cancer revealed three mitochondrial genes potentially associated with poor outcomes of patients. J Transl Med. 2021a;19:273.)
Integrative Multiomics analysis for breast cancer biomarker prediction	Breast cancer	GO enrichment analysis	R and python	transcriptomics, epigenomics, and proteomics data	2832	The Cancer Genome Atlas (TCGA) Molecular Taxonomy of Breast Cancer International Consortium (METABRIC)	(*Fan et al., 2022*Fan Y, Kao C, Yang F, Wang F, Yin G, Wang Y, et al. Integrated Multi-Omics Analysis Model to Identify Biomarkers Associated with Prognosis of Breast Cancer. Front Oncol. 2022;12:899900.)
Multiomics analysis for lung cancer biomarker prediction	Lung cancer	GO enrichment analysis and KEGG analysis	Python	Genomics	189	The Cancer Genome Atlas (TCGA)	(*Li et al., 2022*Li W, Liu B, Wang W, Sun C, Che J, Yuan X, Zhai C. Lung Cancer Stage Prediction Using Multi-Omics Data. Comput Math Methods Med. 2022;2022:2279044.)

Approach	Type of biomarkers	Algorithm	Data source	Results	Ref.
Diagnostic biomarkers	Transcriptomic biomarker	Random Forest	Transcriptomics data from blood samples of patients with a particular disease and healthy controls.	Random Forest was used to predict disease status based on the expression levels of transcripts. Several transcripts were found to be associated with disease status, and these transcripts were used as potential transcriptomic biomarkers for disease diagnosis	(*Huseby et al., 2022*Huseby CJ, Delvaux E, Brokaw DL, Coleman PD. Blood Transcript Biomarkers Selected by Machine Learning Algorithm Classify Neurodegenerative Diseases including Alzheimer’s Disease. Biomolecules. 2022;12(11):1592.)
Diagnostic biomarkers	Metabolomic biomarker	K-Nearest Neighbors	Metabolomics data from urine samples of patients with a particular disease and healthy controls.	K-Nearest Neighbours was used to predict disease status based on the levels of metabolites. Several metabolites were found to be associated with disease status, and these metabolites were used as potential metabolomic biomarkers for disease diagnosis	(*Gowda et al., 2008*Gowda GA, Zhang S, Gu H, Asiago V, Shanaiah N, Raftery D. Metabolomics-based methods for early disease diagnostics. Expert Rev Mol Diagn. 2008;8(5):617-633.)
Diagnostic biomarkers	Proteomic biomarker	Support Vector Machines	Proteomics data from serum samples of patients with a particular disease and healthy controls.	Support Vector Machines was used to predict disease status depending on the number of proteins. Several proteins were found to be associated with disease status, and these proteins were used as potential proteomic biomarkers chemotherapy resistance prediction in small cell lung cancer.	(*Han et al., 2012*Han M, Dai J, Zhang Y, Lin Q, Jiang M, Xu X, et al. Support vector machines coupled with proteomics approaches for detecting biomarkers predicting chemotherapy resistance in small cell lung cancer. Oncology Reports. 2012;28:2233-2238.)
Prognostic biomarkers	Transcriptomic biomarker	Artificial Neural Network	Transcriptomics data from tissue samples of patients with a particular disease and healthy controls.	Artificial Neural Network was used to predict disease progression based on the expression levels of transcripts. Several transcripts were found to be associated with breast cancer progression, and these transcripts were used as potential transcriptomic biomarkers for breast cancer prognosis.	(*Chen et al., 2021*Chen X, Chen DG, Zhao Z, Balko JM, Chen J. Artificial image objects for classification of breast cancer biomarkers with transcriptome sequencing data and convolutional neural network algorithms. Breast Cancer Res. 2021;23(1):96.)
	Metabolomic biomarker	Logistic Regression	Metabolomics data from blood samples of Alzheimer’s patients and healthy controls.	The study found that levels of certain metabolites, such as choline and lactate, were significantly different between Alzheimer’s patients and healthy controls. Logistic Regression was used to build a model that could determine Alzheimer’s disease based on these metabolites. The model was able to accurately predict Alzheimer’s disease with high sensitivity and specificity.	(*Wang et al., 2021*Wang YY, Sun YP, Luo YM, Peng DH, Li X, Yang BY, et al. Biomarkers for the Clinical Diagnosis of Alzheimer’s Disease: Metabolomics Analysis of Brain Tissue and Blood. Front Pharmacol. 2021;12.)
	Transcriptomic biomarker	Random Forest	Transcriptomics data from blood samples of patients with breast cancer and healthy controls.	The study found that levels of certain transcripts were significantly different between breast cancer patients and healthy controls. Random Forest was used to build a model that could predict breast cancer prognosis based on these transcripts. This was able to accurately determine the prognosis of breast cancer accurately.	(Zare, Postovit, Githaka, 2021Zare A, Postovit LM, Githaka JM. Robust inflammatory breast cancer gene signature using nonparametric random forest analysis. Breast Cancer Res. 2021;23(1):92.)
Predictive biomarkers	Transcriptomic biomarker	Random Forest	Transcriptomics data from biopsy samples of patients with a particular disease and healthy controls.	Random Forest was used to predict treatment response based on the expression levels of genes. Several genes were found to be associated with treatment response, and these genes were used as potential transcriptomic biomarkers for rheumatoid arthritis.	(*Rychkov et al., 2021*Rychkov D, Neely J, Oskotsky T, Yu S, Perlmutter N, Nititham J, et al. Cross-Tissue Transcriptomic Analysis Leveraging Machine Learning Approaches Identifies New Biomarkers for Rheumatoid Arthritis. Front Immunol. 2021;12:638066.)
	Metabolomic biomarker	Support Vector Machines (SVM)	Metabolomics data from blood samples of patients with a particular disease and healthy controls.	SVM was used to predict treatment response based on the levels of metabolites. Several metabolites were found to be associated with treatment response, and these metabolites were used as potential metabolomic biomarkers for treatment prediction.	(*Liu et al., 2022*Liu J, Huang L, Shi X, Gu C, Xu H, Liu S. Clinical parameters and metabolomic biomarkers that predict in-hospital outcomes in patients with ST-segment elevated myocardial infarctions. Frontiers Physiol. 2022; 12:820240.)
	Epigenetic biomarker	Logistic Regression	Data of DNA methylation from blood mononuclear cells of patients with a particular disease and healthy controls.	Logistic Regression was used to predict treatment response based on the levels of DNA methylation. Several regions of DNA methylation were found to be associated with treatment response, and these regions were used as potential epigenetic biomarkers for treatment prediction.	(*Cappozzo et al., 2022*Cappozzo A, McCrory C, Robinson O, Freni Sterrantino A, Sacerdote C, Krogh V, et al. A blood DNA methylation biomarker for predicting short-term risk of cardiovascular events. Clinical Epigenetics. 2022;14(1):121.)

Approach	Type of biomarkers	Algorithm	Data source	Results	Ref.
Diagnostic biomarkers	miRNA expression levels using RNA-seq data	Machine learning algorithms.	Biochimica et Biophysica Acta (BBA) - Molecular Basis of Disease	miRNAs were found to have diagnostic potential for several types of cancers, such as liver and lung cancer	(*Pandey et al., 2021*Pandey M, Mukhopadhyay A, Sharawat SK, Kumar S. Role of microRNAs in regulating cell proliferation, metastasis and chemoresistance and their applications as cancer biomarkers in small cell lung cancer. Biochimica et Biophysica Acta (BBA)-Reviews on Cancer. 2021;1876(1):188552.)
	DNA methylation patterns using genome-wide	Machine learning algorithms.	Epigenetics	DNA methylation was found to have diagnostic potential for several types of cancers, such as colon and breast cancer	(Ding, Chen, Shi, 2019Ding W, Chen G, Shi T. Integrative analysis identifies potential DNA methylation biomarkers for pan-cancer diagnosis and prognosis. Epigenetics. 2019;14(1):67-80.)
	Multiple reaction monitoring-mass spectrometry	Machine learning algorithms	Scientific Reports	Proteins were found to have diagnostic potential for several diseases, such as Alzheimer’s disease	(*Kim et al., 2022*Kim Y, Kim J, Son M, Lee J, Yeo I, Choi KY, et al. Plasma protein biomarker model for screening Alzheimer disease using multiple reaction monitoring-mass spectrometry. Sci Rep. 2022;12(1):1282.)
Prognostic biomarkers	miRNA expression using miRNA microarray data	Machine learning algorithms	Cancer Res 2008	miRNA expression was found to have prognostic potential for Human epithelial ovarian cancer.	(*Yang et al., 2008*Yang N, Kaur S, Volinia S, Greshock J, Lassus H, Hasegawa H, et al. MicroRNA microarray identifies Let-7i as a novel biomarker and therapeutic target in human epithelial ovarian cancer. Cancer Res. 2008;68(24):10307-14.)
	Gene methylation using DNA methylation microarray data	Statistical algorithms	Ageing	Gene methylation was found to have prognostic potential for colorectal cancer survival	(*Fu et al., 2020*Fu B, Du C, Wu Z, Li M, Zhao Y, Liu X, et al. Analysis of DNA methylation-driven genes for predicting the prognosis of patients with colorectal cancer. Aging (Albany NY). 2020;12(22):22814-22839.)
	Tumor infiltrate lymphocytes (TILs) using immunohistochemistry	Image analysis algorithms	Frontiers In immunology	TILs were found to have prognostic potential for cutaneous melanoma patient survival	(*Maibach et al., 2020*Maibach F, Sadozai H, Seyed Jafari SM, Hunger RE, Schenk M. Tumor-Infiltrating Lymphocytes and Their Prognostic Value in Cutaneous Melanoma. Front Immunol. 2020;11:2105.)
Predictive biomarkers	Gene expression profiling using RNA-seq data	Machine learning algorithms	Cancers 2021	Gene expression profiling was found to have predictive potential for non-small cell lung cancer patients	(*Hijazo-Pechero et al., 2021*Hijazo-Pechero S, Alay A, Marín R, Vilariño N, Muñoz-Pinedo C, Villanueva A, et al. Gene Expression Profiling as a Potential Tool for Precision Oncology in Non-Small Cell Lung Cancer. Cancers. 2021;13(19):4734.)
	Allele-specific copy number analysis of tumors	Statistical algorithms	Scientific Report	Genomic CNV may be a novel prognostic biomarker for WHO grade IV glioma patient outcomes.	(*Buchwald et al., 2020*Buchwald ZS, Tian S, Rossi M, Smith GH, Switchenko J, Hauenstein JE, et al. Genomic copy number variation correlates with survival outcomes in WHO grade IV glioma. Sci Rep. 2020;10(1):7355.)
	Circulating tumor DNA (ctDNA) using NGS data	machine learning algorithms	Life (Basel)	ctDNA was found to have predictive potential for drug response in cancer patients.	(*Lin et al., 2021*Lin C, Liu X, Zheng B, Ke R, Tzeng CM. Liquid Biopsy, ctDNA Diagnosis through NGS. Life (Basel). 2021;11(9):890.)

Universidade de São Paulo, Faculdade de Ciências Farmacêuticas Av. Prof. Lineu Prestes, n. 580, 05508-000 S. Paulo/SP Brasil, Tel.: (55 11) 3091-3824 - São Paulo - SP - Brazil
E-mail: bjps@usp.br

Acompanhe os números deste periódico no seu leitor de RSS

[1] *Correspondence: A. S. Raikar. Research Scholar. Dept. of Pharmaceutics. PES Rajaram and Tarabai Bandekar College of Pharmacy. Email: amisharaikarofficial@gmail.com

\Feedforward artificial neural network	It has data flowing in one direction through layers of neurons, with the output being achieved at the end.
Recurrent neural networks	These networks use time-series data or sequences and have the ability to “remember” previous events in the sequence.
Long/Short-term Memory	This is an advanced form of RNN that can remember events from several layers ago through memory cells.
Convolutional neural networks	These networks are commonly used in image recognition and use several layers to filter different parts of an image.
Generative Adversarial Networks	These networks involve two neural networks competing with each other to improve accuracy, where one network creates examples and the other network tries to prove them true or false.