Publications

Pre-prints

OnSIDES (ON-label SIDE effectS resource) Database : Extracting Adverse Drug Events from Drug Labels using Natural Language Processing Models Yutaro Tanaka, Hsin Yi Chen, Pietro Belloni, Undina Gisladottir, Jenna Kefeli, Jason Patterson, Apoorva Srinivasan, Michael Zietz, Gaurav Sirdeshmukh, Jacob Berkowitz, Kathleen LaRow Brown, Nicholas P. Tatonetti medRxiv 2024.03.22.24304724; doi: https://doi.org/10.1101/2024.03.22.24304724
Large Language Models for Granularized Barrett's Esophagus Diagnosis Classification J Kefeli, A Soroush, CJ Diamond, HM Zylberberg, B May, JA Abrams, ... arXiv preprint arXiv:2308.08660 https://arxiv.org/abs/2308.08660
Transfer learning from simulations improves the classification of OCT images of glandular epithelia Sassan Ostvar, Han Truong, Elisabeth R. Silver, Charles J. Lightdale, Chin Hur, Nicholas P. Tatonetti bioRxiv 2020.10.26.355180; doi: https://doi.org/10.1101/2020.10.26.355180
Alexandre Yahi, Nicholas P Tatonetti. Simulating drug effects on blood glucose laboratory test time series with a conditional WGAN medRxiv 2020.07.19.20157321; doi: https://doi.org/10.1101/2020.07.19.20157321
Vijendra Ramlall, Kayla M. Quinnies, Rami Vanguri, Tal Lorberbaum, David B. Goldstein, Nicholas P. Tatonetti. Predicting the genetic ancestry of 2.6 million New York City patients using clinical data bioRxiv 768440; doi: https://doi.org/10.1101/768440
Phyllis M. Thangaraj, Nicholas P. Tatonetti. Medical data and machine learning improve power of stroke genome-wide association studies. bioRxiv 2020.01.22.915397; doi: https://doi.org/10.1101/2020.01.22.915397

Peer Reviewed

Kefeli, J., Berkowitz, J., Acitores Cortina, J.M. et al. Generalizable and automated classification of TNM stage from pathology reports with external validation. Nature Communications 15, 8916 (2024). https://doi.org/10.1038/s41467-024-53190-9
Patterson J, Tatonetti N, KG-LIME: predicting individualized risk of adverse drug events for multiple sclerosis disease-modifying therapy, Journal of the American Medical Informatics Association, Volume 31, Issue 8, August 2024, Pages 1693–1703, https://doi.org/10.1093/jamia/ocae155
Kefeli J, Tatonetti NP. TCGA-Reports: A machine-readable pathology report resource for benchmarking text-based AI models. Patterns. 2024, Patterns 5, 100933 doi.org/10.1016/j.patter.2024.100933
Estimating the heritability of SARS-CoV-2 susceptibility and COVID-19 severity KLR Brown, V Ramlall, M Zietz, U Gisladottir, NP Tatonetti Nature Communications 15 (1), 367 doi.org/10.1038/s41467-023-44250-7
Using machine learning probabilities to identify effects of COVID-19 V Ramlall, U Gisladottir, J Kefeli, Y Tanaka, B May, N Tatonetti 2023, Patterns 4, 100889 doi.org/10.1016/j.patter.2023.100889
Joseph D Romano, Hai Li, Tanya Napolitano, Ronald Realubit, Charles Karan, Mandë Holford, Nicholas P Tatonetti Toxins 15 (7), 451 doi.org/10.3390/toxins15070451
Structured deep embedding model to generate composite clinical indices from electronic health records for early detection of pancreatic cancer J Park, MG Artin, KE Lee, BL May, M Park, C Hur, NP Tatonetti Patterns 4 (1) doi.org/10.1016/j.patter.2022.100636
Giangreco NP, Tatonetti NP. A database of pediatric drug effects to evaluate ontogenic mechanisms from child growth and development. Med. 2022;3(8):579-595.e7. doi:10.1016/j.medj.2022.06.001
Giangreco NP, Tatonetti NP. Evaluating risk detection methods to uncover ontogenic-mediated adverse drug effect mechanisms in children. BioData Min. 2021;14(1):34. Published 2021 Jul 22. doi:10.1186/s13040-021-00264-9
Thangaraj PM, Kummer BR, Lorberbaum T, Elkind MSV, Tatonetti NP. Comparative analysis, applications, and interpretation of electronic health record-based stroke phenotyping methods. BioData Min. 2020;13(1):21. Published 2020 Dec 7. doi:10.1186/s13040-020-00230-x
Chandak P, Tatonetti NP. Using Machine Learning to Identify Adverse Drug Effects Posing Increased Risk to Women. Patterns (N Y). 2020;1(7):100108. doi:10.1016/j.patter.2020.100108
Butler DJ, Mozsary C, Meydan C, et al. Shotgun Transcriptome and Isothermal Profiling of SARS-CoV-2 Infection Reveals Unique Host Responses, Viral Diversification, and Drug Interactions. Preprint. bioRxiv. 2020;2020.04.20.048066. Published 2020 May 1. doi:10.1101/2020.04.20.048066
Zietz M, Zucker J, Tatonetti NP. Associations between blood type and COVID-19 infection, intubation, and death. Nat Commun. 2020;11(1):5761. Published 2020 Nov 13. doi:10.1038/s41467-020-19623-x

“Reviewing medical records for 7,770 people who tested positive for the coronavirus, Dr. Tatonetti and a graduate student, Michael Zietz, found that people with Type A blood were at a somewhat lower risk of being placed on ventilators. People who were Type AB were at a higher risk, but the scientists cautioned that this result might not be reliable because there were so few patients with that blood type in their analysis.”
Covid-19 Risk Doesn’t Depend (Much) on Blood Type, New Studies Find. July 15, 2020, Continue Reading

Ramlall V, Thangaraj PM, Meydan C, et al. Immune complement and coagulation dysfunction in adverse outcomes of SARS-CoV-2 infection. Nat Med. 2020;26(10):1609-1615. doi:10.1038/s41591-020-1021-2
Basile AO, Yahi A, Tatonetti NP. Artificial Intelligence for Drug Toxicity and Safety. Trends Pharmacol Sci. 2019;40(9):624-635. doi:10.1016/j.tips.2019.07.005
Glicksberg BS, Oskotsky B, Thangaraj PM, et al. PatientExploreR: an extensible application for dynamic visualization of patient clinical history from electronic health records in the OMOP common data model. Bioinformatics. 2019;35(21):4515-4518. doi:10.1093/bioinformatics/btz409
Romano JD, Tatonetti NP. Informatics and Computational Methods in Natural Product Drug Discovery: A Review and Perspectives. Front Genet. 2019;10:368. Published 2019 Apr 30. doi:10.3389/fgene.2019.00368
Boland MR, Tatonetti NP. Attention Deficit-Hyperactivity Disorder and Month of School Enrollment. N Engl J Med. 2019;380(7):692-693. doi:10.1056/NEJMc1817539
Ta CN, Dumontier M, Hripcsak G, Tatonetti NP, Weng C. Columbia Open Health Data, clinical concept prevalence and co-occurrence from electronic health records. Sci Data. 2018;5:180273. Published 2018 Nov 27. doi:10.1038/sdata.2018.273
Shameer K, Perez-Rodriguez MM, Bachar R, et al. Pharmacological risk factors associated with hospital readmission rates in a psychiatric cohort identified using prescriptome data mining. BMC Med Inform Decis Mak. 2018;18(Suppl 3):79. Published 2018 Sep 14. doi:10.1186/s12911-018-0653-3
Hao Y, Quinnies K, Realubit R, Karan C, Tatonetti NP. Tissue-Specific Analysis of Pharmacological Pathways. CPT Pharmacometrics Syst Pharmacol. 2018;7(7):453-463. doi:10.1002/psp4.12305
Faye AS, Polubriaginof F, Green PHR, Vawdrey DK, Tatonetti N, Lebwohl B. Low Rates of Screening for Celiac Disease Among Family Members. Clin Gastroenterol Hepatol. 2019;17(3):463-468. doi:10.1016/j.cgh.2018.06.016
Polubriaginof FCG, Vanguri R, Quinnies K, et al. Disease Heritability Inferred from Familial Relationships Reported in Medical Records. Cell. 2018;173(7):1692-1704.e11. doi:10.1016/j.cell.2018.04.032

“The team was pleased to find that height, obesity, diabetes, acne and sickle cell anemia all came out within a few percentage points of more traditional research studies. The data set also offered potential heritability data for about 400 items that had not previously been examined this way, including sinus infections, tooth decay, irregular menstruation and thyroid disorders.”
‘Will You Be My Emergency Contact?’ Takes On a Whole New Meaning May 17, 2018, Continue Reading

“Hoping to explore the genetics of drug reactions, graduate student Fernanda Polubriaginof and others working in the lab of biomedical informatics researcher Nicholas Tatonetti at Columbia University wanted to determine whether patients at the school’s affiliated NewYork-Presbyterian Hospital were related.”
Family trees hidden in medical records could predict your disease risk, May 17, 2018, Continue Reading

“Going to the hospital means filling out forms, and somewhere on that form is a place to list an emergency contact, with name, address, phone number, and relationship just in case. A study published Thursday in Cell found a novel use for that information.”
Researchers mine emergency contacts to discover genetic clues to disease, May 17, 2018, Continue Reading

“Because many people use relatives as their contact person and that contact might list another relative or next-of-kin information on their own forms, researchers from Columbia University and elsewhere were able to generate 223,000 family trees connecting blood relatives using electronic health record data from 3.5 million patients at three medical centers in New York.”
Family Ties, May 17, 2018, Continue Reading

Tatonetti NP. The Next Generation of Drug Safety Science: Coupling Detection, Corroboration, and Validation to Discover Novel Drug Effects and Drug-Drug Interactions. Clin Pharmacol Ther. 2018;103(2):177-179. doi:10.1002/cpt.949
Alexandre Yahi, Rami Vanguri, Noemie Elhadad, Nicholas P Tatonetti, Generative Adversarial Networks for Electronic Health Records: A Framework for Exploring and Evaluating Methods for Predicting Drug-Induced Laboratory Test Trajectories 31st Conference on Neural Information Processing Systems (NIPS 2017) December 2017
Macesic N, Polubriaginof F, Tatonetti NP. Machine learning: novel bioinformatics approaches for combating antimicrobial resistance. Curr Opin Infect Dis. 2017;30(6):511-517. doi:10.1097/QCO.0000000000000406
Moskovitch R, Polubriaginof F, Weiss A, Ryan P, Tatonetti N. Procedure prediction from symbolic Electronic Health Records via time intervals analytics. J Biomed Inform. 2017;75:70-82. doi:10.1016/j.jbi.2017.07.018
Moskovitch R, Polubriaginof F, Weiss A, Ryan P, Tatonetti N. Procedure prediction from symbolic Electronic Health Records via time intervals analytics. J Biomed Inform. 2017;75:70-82. doi:10.1016/j.jbi.2017.07.018
Boland MR, Polubriaginof F, Tatonetti NP. Development of A Machine Learning Algorithm to Classify Drugs Of Unknown Fetal Effect. Sci Rep. 2017;7(1):12839. Published 2017 Oct 9. doi:10.1038/s41598-017-12943-x
Boland MR, Parhi P, Li L, et al. Uncovering exposures responsible for birth season - disease effects: a global study. J Am Med Inform Assoc. 2018;25(3):275-288. doi:10.1093/jamia/ocx105
Boland MR, Parhi P, Gentine P, Tatonetti NP. Climate Classification is an Important Factor in Assessing Quality-of-Care Across Hospitals. Sci Rep. 2017;7(1):4948. Published 2017 Jul 10. doi:10.1038/s41598-017-04708-3
Moskovitch R, Choi H, Hripcsak G, Tatonetti N. Prognosis of Clinical Outcomes with Temporal Patterns and Experiences with One Class Feature Selection. IEEE/ACM Trans Comput Biol Bioinform. 2017;14(3):555-563. doi:10.1109/TCBB.2016.2591539
Boland MR, Karczewski KJ, Tatonetti NP. Ten Simple Rules to Enable Multi-site Collaborations through Data Sharing. PLoS Comput Biol. 2017;13(1):e1005278. Published 2017 Jan 19. doi:10.1371/journal.pcbi.1005278
Hao Y, Tatonetti NP. Predicting G protein-coupled receptor downstream signaling by tissue expression. Bioinformatics. 2016;32(22):3435-3443. doi:10.1093/bioinformatics/btw510
Boland MR, Tatonetti NP. Investigation of 7-dehydrocholesterol reductase pathway to elucidate off-target prenatal effects of pharmaceuticals: a systematic review. Pharmacogenomics J. 2016;16(5):411-429. doi:10.1038/tpj.2016.48
Lorberbaum T, Sampson KJ, Chang JB, et al. Coupling Data Mining and Laboratory Experiments to Discover Drug Interactions Causing QT Prolongation. J Am Coll Cardiol. 2016;68(16):1756-1764. doi:10.1016/j.jacc.2016.07.761

“By examining big data in entirely new ways, the team uncovered several drug combinations associated with increased risk of a potentially fatal heart arrhythmia. One pair included the popular antibiotic ceftriaxone and the heartburn medication lansoprazole, a former blockbuster drug best known by the brand name Prevacid.”
Scientists see progress in identifying deadly drug interactions, October 10, 2016, Continue Reading

“Millions of Americans take more than one prescription drug, and often times doctors don’t know which drugs, when combined, can cause serious illness or death. Sadly, such warnings come only after the damage is done, when enough clear reports of adverse reactions begin to emerge. Now, scientists at Columbia University in New York have harnessed the power of data science to identify two common prescription drugs that, if mixed, can have deadly consequences.”
Deadly Mixture: Scientists Uncover Harmful Drug Interactions, October 10, 2016, Continue Reading

“The research, which used a combination of adverse event data-mining, electronic health record analysis and laboratory experiments, found that patients who received both drugs together were more likely to experience drug-induced long QT syndrome (LQTS), a potentially fatal condition, than those who received either drug alone.”
Study highlights drug-drug interaction risk from two commonly used drugs, October 12, 2016, Continue Reading

“Investigators at Columbia University discovered that adverse event reports and electronic health records, along with targeted tests, were effective in confirming the challenging-to-predict interactions – pinpointing the peril of one specific combination therapy.”
Data mining, lab testing approach reveals ceftriaxone-lansoprazole interaction, October 20, 2016,Continue Reading

Chang JB, Quinnies KM, Realubit R, Karan C, Rand JH, Tatonetti NP. A novel, rapid method to compare the therapeutic windows of oral anticoagulants using the Hill coefficient. Sci Rep. 2016;6:29387. Published 2016 Jul 21. doi:10.1038/srep29387
Lorberbaum T, Sampson KJ, Woosley RL, Kass RS, Tatonetti NP. An Integrative Data Science Pipeline to Identify Novel Drug Interactions that Prolong the QT Interval. Drug Saf. 2016;39(5):433-441. doi:10.1007/s40264-016-0393-1

“The experiment began with thousands of patient files, millions of prescription orders, billions of clinical measurements and a single question: Could big data be used to discover deadly drug combinations?”
Hunt for dangerous drug interactions reveals strategy that can save lives, February 11, 2016, Continue Reading

Romano JD, Tatonetti NP. VenomKB, a new knowledge base for facilitating the validation of putative venom therapies. Sci Data. 2015;2:150065. Published 2015 Nov 24. doi:10.1038/sdata.2015.65

“The bite of a poisonous snake, scorpion or other venomous creature could very well kill you, but it also might be able to heal certain medical conditions like cancer, diabetes and heart failure”
Animal Venom Database Could Be Boon To Drug Development , November 29, 2015, Continue Reading

“Sometimes what should hurt you, helps you, and what sounds like folklore turns out to be really science. The saliva of the Gila Monster, an orange speckled lizard native to the American Southwest, is poisonous – but it also can be used to effectively treat Type 2 diabetes…”VenomKB, a Therapeutic Natural Toxins Database, Makes Folklore Into Science , December 8, 2015,Continue Reading

Jacunski A, Dixon SJ, Tatonetti NP. Connectivity Homology Enables Inter-Species Network Models of Synthetic Lethality. PLoS Comput Biol. 2015;11(10):e1004506. Published 2015 Oct 9. doi:10.1371/journal.pcbi.1004506
Boland MR, Shahn Z, Madigan D, Hripcsak G, Tatonetti NP. Birth month affects lifetime disease risk: a phenome-wide method. J Am Med Inform Assoc. 2015;22(5):1042-1053. doi:10.1093/jamia/ocv046

“By delving into the extensive database of patients seen at Columbia Medical Center over 14 years, beginning in 2000, Tatonetti and his team did a first-of-its kind look at whether birth month has anything to do with disease risk.”
See What Diseases You’re at Risk For Based on Your Birth Month, June 8, 2015, Continue Reading

“Mary Regina Boland, Nicholas Tatonetti and other researchers at the Columbia University Department of Medicine examined records for an incredible 1.75 million patients born between 1900 and 2000 who had been treated at Columbia University Medical Center. Using statistical analysis, they combed through 1,688 different diseases and found 55 that had a correlation with birth month.”
Scientists have discovered how the month you’re born matters for your health, June 15, 2015, Continue Reading

“The Columbia researchers also found that one in 675 occurrences of ADHD (attention deficit hyperactivity disorder) could relate to being born in New York in November, which matches a Swedish study showing the highest rates of ADHD in November babies.”
Could your birth month indicate your risk of disease?, June 9, 2015, Continue Reading

“The results of this study should be interpreted carefully, researchers said. For one, the study only examined data collected from one hospital in New York City; therefore, health data may be skewed due to New York’s climate. Furthermore, sick patients tend to be overrepresented in electronic health records, because, well, they are visiting the doctor more often.”
Your Birth Month Influences Your Risk for Diseases, June 9, 2015, Continue Reading

“Columbia University scientists compiled the birthdays and medical records of patients from New York City databases and found that people with May birthdays may have the healthiest outcomes, while people born in October might be at the highest risk for certain diseases.”
Birth month may correlate to some diseases (bad news, October), June 8, 2015, Continue Reading

[VIDEO] “Some people swear by the predictive powers of zodiac signs. And while there’s no scientific evidence to support Capricorns having better luck with Scorpios, this guy says the month you’re born in could make a difference in your future health.”
The study says there are at least 55 diseases that are significantly dependent on birth month., June 10, 2015, Watch the Video

Lorberbaum T, Nasir M, Keiser MJ, Vilar S, Hripcsak G, Tatonetti NP. Systems pharmacology augments drug safety surveillance. Clin Pharmacol Ther. 2015;97(2):151-158. doi:10.1002/cpt.2
Vilar S, Uriarte E, Santana L, et al. Similarity-based modeling in large-scale prediction of drug-drug interactions. Nat Protoc. 2014;9(9):2147-2163. doi:10.1038/nprot.2014.151
Karczewski KJ, Snyder M, Altman RB, Tatonetti NP. Coherent functional modules improve transcription factor target identification, cooperativity prediction, and disease association. PLoS Genet. 2014;10(2):e1004122. Published 2014 Feb 6. doi:10.1371/journal.pgen.1004122
Jacunski A, Tatonetti NP. Connecting the dots: applications of network medicine in pharmacology and disease. Clin Pharmacol Ther. 2013;94(6):659-669. doi:10.1038/clpt.2013.168
Sun X, Vilar S, Tatonetti NP. High-throughput methods for combinatorial drug discovery. Sci Transl Med. 2013;5(205):205rv1. doi:10.1126/scitranslmed.3006667
White RW, Tatonetti NP, Shah NH, Altman RB, Horvitz E. Web-scale pharmacovigilance: listening to signals from the crowd. J Am Med Inform Assoc. 2013;20(3):404-408. doi:10.1136/amiajnl-2012-001482

Listen to Dr. Eric Horvitz discuss our work with Ira Flatow on NPR’s Science Friday.
Improving Healthcare, One Search at a Time, March 15, 2013, source

“…the Stanford and Columbia University joint research team sifted though 6 million users’ internet search queries (which you’ll be uncomfortable to know, are forever saved in web search logs) and looked for searches that related to the antidepressant paroxetine and the cholesterol-lowering drug pravastatin.”
Your Google Searches Can Uncover Drug Side Effects Faster Than the FDA, March 6, 2013, source

“Much like Google Flu Trends reveals influenza outbreaks by tracking flu-related search terms, search queries about drug combinations and possible side effects-say, “paroxetine,” “pravastatin,” and “hyperglycemia”-might enable researchers to identify unanticipated downsides.”
Should You Mix Those Two Drugs? Ask Dr. Google, March 6, 2013, source

“Using automated software tools to examine queries by six million Internet users taken from Web search logs in 2010, the researchers looked for searches relating to an antidepressant, paroxetine, and a cholesterol lowering drug, pravastatin. They were able to find evidence that the combination of the two drugs caused high blood sugar.”
Unreported Side Effects of Drugs Are Found Using Internet Search Data, Study Finds, March 6, 2013, source

Tatonetti NP, Ye PP, Daneshjou R, Altman RB. Data-driven prediction of drug effects and interactions. Sci Transl Med. 2012;4(125):125ra31. doi:10.1126/scitranslmed.3003377

“An algorithm designed by US scientists to trawl through a plethora of drug interactions has yielded thousands of previously unknown side effects caused by taking drugs in combination”
Drug data reveal sneaky side effects, March 14, 2012, source

“… researchers have shown thousands of previously unknown side effects caused when some drugs are taken in combination …”
Good Apart, Bad Together, March 15, 2012, source

“There’s always that part at the end of drug commercials that goes something like: if you develop sausage fingers, webbed feet, or a three-week erection, call your doctor! But as exhaustive as those auctioneer-style lists sound, they barely scratch the surface when it comes to the side effects people are actually experiencing.”
Drugs Cause About Five Times More Side Effects Than We Realized, March 15, 2012, source

“It’s a funny thing about clinical trials: they’re set up so that all the subjects taking the drug in question are totally healthy in every other respect and on no other medications. It’s the only way to see if the drug has an effect, but as soon as people start taking it in the real world, well… let’s just say that most patients don’t measure up to the pristine condition of clinical subjects.”
Analysis of Drug Database Reveals Thousands of Potentially Dangerous Interactions, March 15, 2012, source

“Un grupo de investigadores ha disenado metodos computacionales para predecir las interacciones entre medicamentos y sus consecuencias adversas …”
Predicen con algoritmos los efectos secundarios de mezclar farmacos, March 15, 2012, source

Karczewski KJ, Tirrell RP, Cordero P, et al. Interpretome: a freely available, modular, and secure personal genome interpretation engine. Pac Symp Biocomput. 2012;339-350.
Tatonetti NP, Fernald GH, Altman RB. A novel signal detection algorithm for identifying hidden drug-drug interactions in adverse event reports. J Am Med Inform Assoc. 2012;19(1):79-85. doi:10.1136/amiajnl-2011-000214
Karczewski KJ, Tatonetti NP, Landt SG, et al. Cooperative transcription factor associations discovered using regulatory variation. Proc Natl Acad Sci U S A. 2011;108(32):13353-13358. doi:10.1073/pnas.1103105108
Tatonetti NP, Denny JC, Murphy SN, et al. Detecting drug interactions from adverse-event reports: interaction between paroxetine and pravastatin increases blood glucose levels [published correction appears in Clin Pharmacol Ther. 2011 Sep;90(3):480. Tsau, P S [corrected to Tsao, P S]]. Clin Pharmacol Ther. 2011;90(1):133-142. doi:10.1038/clpt.2011.83

“Dr. Tatonetti devised an algorithm to look for pairs of drugs that, taken together, cause a side effect not associated with either drug alone. One pairing popped up when he used his new software to search the Food and Drug Administration’s database of adverse drug reports: Paxil, a widely used antidepressant, and Pravastatin, a cholesterol-lowering drug. …”
Mining Electronic Records for Revealing Health Data, January 14, 2013, source

“Up to 1 million patients in the United States may be taking 2 medications that can lead to unexpected increases in blood glucose levels when used simultaneously. Data mining techniques have revealed that the combination of the antidepressant paroxetine and the cholesterol-lowering medication pravastatin may cause this adverse effect …”
Data Mining Approach Shows Promise in Detecting Unexpected Drug Interactions, July 13, 2011, source

“Researchers mined FDA’s Adverse Event Reporting System (AERS) for reports of side effects involving glucose homeostasis.”
Data-mining uncovers hyperglycemic drug-drug interaction between paroxetine and pravastatin, August 15, 2011, source

“Americans have been led to believe — by their doctors, by advertisers and by the pharmaceutical industry — that there is a pill to cure just about anything that ails them.”
Are you taking too many meds?, May 31, 2011, source

“The side effect … was unexpected, and its discovery illustrates the power of electronic health records to help bring to light previously unknown problems with medical treatments.”
Study warns on use of 2 common drugs, May 28, 2011, source

“[By] combin[ing] a list of drugs known to affect pathways involved in diabetes, and then mined AERS for side effects associated with these drugs. Then they set their algorithms loose on AERS to find combinations of drugs that produced the same constellation of side effects, thinking that these might also affect pathways involved in diabetes.”
Common drug combo increases diabetes risk, May, 2011, source

Tatonetti NP, Liu T, Altman RB. Predicting drug side-effects by chemical systems biology. Genome Biol. 2009;10(9):238. doi:10.1186/gb-2009-10-9-238