Amit Kaushal, MD, PhD is Clinical Associate Professor of Medicine (Stanford-VA) and Adjunct Professor of Bioengineering at Stanford University. Dr. Kaushal's work spans clinical medicine, teaching, research, and industry.
He helped launch Stanford School of Engineering's undergraduate major in Biomedical Computation (bmc.stanford.edu) and has served as long-time director of the major. The major has graduated over 70 students since inception and was recently featured in Nature (https://go.nature.com/2P2UnRu).
His research interests are in utilizing health data in novel and ethical ways to improve the practice of medicine. He is a faculty executive member of Stanford's Partnership for AI-Assisted Care (aicare.stanford.edu). Recently, he has also been working with public health agencies to improve scale and speed of contact tracing for COVID-19.
He has previously held executive and advisory roles at startups working at the interface of technology and healthcare.
He continues to practice as an academic hospitalist.
Dr. Kaushal completed his BS (Biomedical Computation), MD, PhD (Biomedical Informatics), and residency training at Stanford. He is board-certified in Internal Medicine and Clinical Informatics.
Faculty Affiliate, Institute for Human-Centered Artificial Intelligence (HAI)
Executive Director, Biomedical Computation Major (2007 - Present)
Associate Director, Biomedical Computation Major (2003 - 2007)
Honors & Awards
Recipient, Paul and Daisy Soros Fellowship
Residency, Stanford University, Internal Medicine
PhD, Stanford University, Biomedical Informatics
MD, Stanford University
BS, Stanford University, Biomedical Computation
- AI-Assisted Care
MED 277 (Aut)
Prior Year Courses
- AI-Assisted Care
CS 337, MED 277 (Aut)
- AI-Assisted Care
Geographic Distribution of US Cohorts Used to Train Deep Learning Algorithms.
2020; 324 (12): 1212–13
View details for DOI 10.1001/jama.2020.12067
View details for PubMedID 32960230
- Can Contact Tracing Work At COVID Scale? Health Affairs Blog. 2020
Wiring Minds Successfully applying AI to biomedicine requires innovators trained in contrasting cultures
2019; 576 (7787): S62–S63
View details for Web of Science ID 000513817800002
View details for PubMedID 31853071
Beyond duty hours: leveraging large-scale paging data to monitor resident workload
NPJ DIGITAL MEDICINE
2019; 2: 87
Monitoring and managing resident workload is a cornerstone of policy in graduate medical education, and the duty hours metric is the backbone of current regulations. While the duty hours metric measures hours worked, it does not capture differences in intensity of work completed during those hours, which may independently contribute to fatigue and burnout. Few such metrics exist. Digital data streams generated during the usual course of hospital operations can serve as a novel source of insight into workload intensity by providing high-resolution, minute-by-minute data at the individual level; however, study and use of these data streams for workload monitoring has been limited to date. Paging data is one such data stream. In this work, we analyze over 500,000 pages-two full years of pages in an academic internal medicine residency program-to characterize paging patterns among housestaff. We demonstrate technical feasibility, validity, and utility of paging burden as a metric to provide insight into resident workload beyond duty hours alone, and illustrate a general framework for evaluation and incorporation of novel digital data streams into resident workload monitoring.
View details for DOI 10.1038/s41746-019-0165-2
View details for Web of Science ID 000484610000001
View details for PubMedID 31531394
View details for PubMedCentralID PMC6733865
Repurposing Biomedical Informaticians for COVID-19.
Journal of biomedical informatics
The COVID-19 pandemic is an unprecedented challenge to the biomedical research community at the intersection of great uncertainty due to the novelty of the virus and extremely high stakes due to the large global death count. The global quarantine shut-downs complicated scientific matters because many laboratories were closed down unless they were actively doing COVID-19 related research, making repurposing of activities difficult for many biomedical researchers. Biomedical informaticians, who have been primarily able to continue their research through remote work and video conferencing, have been able to maintain normal activities. In addition to continuing ongoing studies, there has been great grass roots interest in helping in the fight against COVID-19. In this commentary, we describe several projects that arose from this desire to help, and the lessons that the authors learned along the way. We then offer some insights into how these lessons might be applied to make scientific progress be more efficient in future crisis scenarios.
View details for DOI 10.1016/j.jbi.2021.103673
View details for PubMedID 33486067
Ethical issues in using ambient intelligence in health-care settings.
The Lancet. Digital health
Ambient intelligence is increasingly finding applications in health-care settings, such as helping to ensure clinician and patient safety by monitoring staff compliance with clinical best practices or relieving staff of burdensome documentation tasks. Ambient intelligence involves using contactless sensors and contact-based wearable devices embedded in health-care settings to collect data (eg, imaging data of physical spaces, audio data, or body temperature), coupled with machine learning algorithms to efficiently and effectively interpret these data. Despite the promise of ambient intelligence to improve quality of care, the continuous collection of large amounts of sensor data in health-care settings presents ethical challenges, particularly in terms of privacy, data management, bias and fairness, and informed consent. Navigating these ethical issues is crucial not only for the success of individual uses, but for acceptance of the field as a whole.
View details for DOI 10.1016/S2589-7500(20)30275-2
View details for PubMedID 33358138
Inference for longitudinal data with nonignorable nonmonotone missing responses
COMPUTATIONAL STATISTICS & DATA ANALYSIS
2014; 72: 77-91
For the analysis of longitudinal data with nonignorable and nonmonotone missing responses, a full likelihood method often requires intensive computation, especially when there are many follow-up times. The authors propose and explore a Monte Carlo method, based on importance sampling, for approximating the maximum likelihood estimators. The finite-sample properties of the proposed estimators are studied using simulations. An application of the proposed method is also provided using longitudinal data on peptide intensities obtained from a proteomics experiment of trauma patients.
View details for DOI 10.1016/j.csda.2013.10.027
View details for Web of Science ID 000330147000006
View details for PubMedCentralID PMC4243943
Trauma-associated human neutrophil alterations revealed by comparative proteomics profiling
PROTEOMICS CLINICAL APPLICATIONS
2013; 7 (7-8): 571-583
Polymorphonuclear neutrophils (PMNs) play an important role in mediating the innate immune response after severe traumatic injury; however, the cellular proteome response to traumatic condition is still largely unknown.We applied 2D-LC-MS/MS-based shotgun proteomics to perform comparative proteome profiling of human PMNs from severe trauma patients and healthy controls.A total of 197 out of ~2500 proteins (being identified with at least two peptides) were observed with significant abundance changes following the injury. The proteomics data were further compared with transcriptomics data for the same genes obtained from an independent patient cohort. The comparison showed that the protein abundance changes for the majority of proteins were consistent with the mRNA abundance changes in terms of directions of changes. Moreover, increased protein secretion was suggested as one of the mechanisms contributing to the observed discrepancy between protein and mRNA abundance changes. Functional analyses of the altered proteins showed that many of these proteins were involved in immune response, protein biosynthesis, protein transport, NRF2-mediated oxidative stress response, the ubiquitin-proteasome system, and apoptosis pathways.Our data suggest increased neutrophil activation and inhibited neutrophil apoptosis in response to trauma. The study not only reveals an overall picture of functional neutrophil response to trauma at the proteome level, but also provides a rich proteomics data resource of trauma-associated changes in the neutrophil that will be valuable for further studies of the functions of individual proteins in PMNs.
View details for DOI 10.1002/prca.201200109
View details for Web of Science ID 000327792800011
View details for PubMedID 23589343
View details for PubMedCentralID PMC3737403
Determination of Burn Patient Outcome by Large-Scale Quantitative Discovery Proteomics
CRITICAL CARE MEDICINE
2013; 41 (6): 1421-1434
OBJECTIVES:: Emerging proteomics techniques can be used to establish proteomic outcome signatures and to identify candidate biomarkers for survival following traumatic injury. We applied high-resolution liquid chromatography-mass spectrometry and multiplex cytokine analysis to profile the plasma proteome of survivors and nonsurvivors of massive burn injury to determine the proteomic survival signature following a major burn injury. DESIGN:: Proteomic discovery study. SETTING:: Five burn hospitals across the United States. PATIENTS:: Thirty-two burn patients (16 nonsurvivors and 16 survivors), 19-89 years old, were admitted within 96 hours of injury to the participating hospitals with burns covering more than 20% of the total body surface area and required at least one surgical intervention. INTERVENTIONS:: None. MEASUREMENTS AND MAIN RESULTS:: We found differences in circulating levels of 43 proteins involved in the acute-phase response, hepatic signaling, the complement cascade, inflammation, and insulin resistance. Thirty-two of the proteins identified were not previously known to play a role in the response to burn. Interleukin-4, interleukin-8, granulocyte macrophage colony-stimulating factor, monocyte chemotactic protein-1, and β2-microglobulin correlated well with survival and may serve as clinical biomarkers. CONCLUSIONS:: These results demonstrate the utility of these techniques for establishing proteomic survival signatures and for use as a discovery tool to identify candidate biomarkers for survival. This is the first clinical application of a high-throughput, large-scale liquid chromatography-mass spectrometry-based quantitative plasma proteomic approach for biomarker discovery for the prediction of patient outcome following burn, trauma, or critical illness.
View details for DOI 10.1097/CCM.0b013e31827c072e
View details for Web of Science ID 000319269400024
View details for PubMedID 23507713
View details for PubMedCentralID PMC3660437
Genomic responses in mouse models poorly mimic human inflammatory diseases
PROCEEDINGS OF THE NATIONAL ACADEMY OF SCIENCES OF THE UNITED STATES OF AMERICA
2013; 110 (9): 3507-3512
A cornerstone of modern biomedical research is the use of mouse models to explore basic pathophysiological mechanisms, evaluate new therapeutic approaches, and make go or no-go decisions to carry new drug candidates forward into clinical trials. Systematic studies evaluating how well murine models mimic human inflammatory diseases are nonexistent. Here, we show that, although acute inflammatory stresses from different etiologies result in highly similar genomic responses in humans, the responses in corresponding mouse models correlate poorly with the human conditions and also, one another. Among genes changed significantly in humans, the murine orthologs are close to random in matching their human counterparts (e.g., R(2) between 0.0 and 0.1). In addition to improvements in the current animal model systems, our study supports higher priority for translational medical research to focus on the more complex human conditions rather than relying on mouse models to study human inflammatory diseases.
View details for DOI 10.1073/pnas.1222878110
View details for Web of Science ID 000315841900063
View details for PubMedID 23401516
View details for PubMedCentralID PMC3587220
The Mouse Blood-Brain Barrier Transcriptome: A New Resource for Understanding the Development and Function of Brain Endothelial Cells
2010; 5 (10)
The blood-brain barrier (BBB) maintains brain homeostasis and limits the entry of toxins and pathogens into the brain. Despite its importance, little is known about the molecular mechanisms regulating the development and function of this crucial barrier. In this study we have developed methods to highly purify and gene profile endothelial cells from different tissues, and by comparing the transcriptional profile of brain endothelial cells with those purified from the liver and lung, we have generated a comprehensive resource of transcripts that are enriched in the BBB forming endothelial cells of the brain. Through this comparison we have identified novel tight junction proteins, transporters, metabolic enzymes, signaling components, and unknown transcripts whose expression is enriched in central nervous system (CNS) endothelial cells. This analysis has identified that RXRalpha signaling cascade is specifically enriched at the BBB, implicating this pathway in regulating this vital barrier. This dataset provides a resource for understanding CNS endothelial cells and their interaction with neural and hematogenous cells.
View details for DOI 10.1371/journal.pone.0013741
View details for Web of Science ID 000283645300021
View details for PubMedID 21060791
View details for PubMedCentralID PMC2966423
A universal carrier test for the long tail of Mendelian disease
REPRODUCTIVE BIOMEDICINE ONLINE
2010; 21 (4): 537-551
Mendelian disorders are individually rare but collectively common, forming a 'long tail' of genetic disease. A single highly accurate assay for this long tail would allow the scaling up of the Jewish community's successful campaign of population screening for Tay-Sachs disease to the general population, thereby improving millions of lives, greatly benefiting minority health and saving billions of dollars. This need has been addressed by designing a universal carrier test: a non-invasive, saliva-based assay for more than 100 Mendelian diseases across all major population groups. The test has been exhaustively validated with a median of 147 positive and 525 negative samples per variant, demonstrating a multiplex assay whose performance compares favourably with the previous standard of care, namely blood-based single-gene carrier tests. Because the test represents a dramatic reduction in the cost and complexity of large-scale population screening, an end to many preventable genetic diseases is now in sight. Moreover, given that the assay is inexpensive and requires only a saliva sample, it is now increasingly feasible to make carrier testing a routine part of preconception care.
View details for DOI 10.1016/j.rbmo.2010.05.012
View details for Web of Science ID 000283400000017
View details for PubMedID 20729146
Plasma Proteome Response to Severe Burn Injury Revealed by O-18-Labeled "Universal" Reference-Based Quantitative Proteomics
JOURNAL OF PROTEOME RESEARCH
2010; 9 (9): 4779-4789
A burn injury represents one of the most severe forms of human trauma and is responsible for significant mortality worldwide. Here, we present the first quantitative proteomics investigation of the blood plasma proteome response to severe burn injury by comparing the plasma protein concentrations of 10 healthy control subjects with those of 15 severe burn patients at two time-points following the injury. The overall analytical strategy for this work integrated immunoaffinity depletion of the 12 most abundant plasma proteins with cysteinyl-peptide enrichment-based fractionation prior to LC-MS analyses of individual patient samples. Incorporation of an 18O-labeled "universal" reference among the sample sets enabled precise relative quantification across samples. In total, 313 plasma proteins confidently identified with two or more unique peptides were quantified. Following statistical analysis, 110 proteins exhibited significant abundance changes in response to the burn injury. The observed changes in protein concentrations suggest significant inflammatory and hypermetabolic response to the injury, which is supported by the fact that many of the identified proteins are associated with acute phase response signaling, the complement system, and coagulation system pathways. The regulation of approximately 35 proteins observed in this study is in agreement with previous results reported for inflammatory or burn response, but approximately 50 potentially novel proteins previously not known to be associated with burn response or inflammation are also found. Elucidating proteins involved in the response to severe burn injury may reveal novel targets for therapeutic interventions as well as potential predictive biomarkers for patient outcomes such as multiple organ failure.
View details for DOI 10.1021/pr1005026
View details for Web of Science ID 000281443700041
View details for PubMedID 20698492
View details for PubMedCentralID PMC2945297
Knowledge-based analysis of microarrays for the discovery of transcriptional regulation relationships
8th Asia Pacific Bioinformatics Conference
BIOMED CENTRAL LTD. 2010
The large amount of high-throughput genomic data has facilitated the discovery of the regulatory relationships between transcription factors and their target genes. While early methods for discovery of transcriptional regulation relationships from microarray data often focused on the high-throughput experimental data alone, more recent approaches have explored the integration of external knowledge bases of gene interactions.In this work, we develop an algorithm that provides improved performance in the prediction of transcriptional regulatory relationships by supplementing the analysis of microarray data with a new method of integrating information from an existing knowledge base. Using a well-known dataset of yeast microarrays and the Yeast Proteome Database, a comprehensive collection of known information of yeast genes, we show that knowledge-based predictions demonstrate better sensitivity and specificity in inferring new transcriptional interactions than predictions from microarray data alone. We also show that comprehensive, direct and high-quality knowledge bases provide better prediction performance. Comparison of our results with ChIP-chip data and growth fitness data suggests that our predicted genome-wide regulatory pairs in yeast are reasonable candidates for follow-up biological verification.High quality, comprehensive, and direct knowledge bases, when combined with appropriate bioinformatic algorithms, can significantly improve the discovery of gene regulatory relationships from high throughput gene expression data.
View details for DOI 10.1186/1471-2105-11-S1-S8
View details for Web of Science ID 000273969500009
View details for PubMedID 20122245
View details for PubMedCentralID PMC3009543
Shotgun proteomics identifies proteins specific for acute renal transplant rejection
PROTEOMICS CLINICAL APPLICATIONS
2010; 4 (1): 32-47
Acute rejection (AR) remains the primary risk factor for renal transplant outcome; development of non-invasive diagnostic biomarkers for AR is an unmet need.We used shotgun proteomics applying LC-MS/MS and ELISA to analyze a set of 92 urine samples, from patients with AR, stable grafts (STA), proteinuria (NS), and healthy controls.A total of 1446 urinary proteins (UP) were identified along with a number of nonspecific proteinuria-specific, renal transplantation specific and AR-specific proteins. Relative abundance of identified UP was measured by protein-level spectral counts adopting a weighted fold-change statistic, assigning increased weight for more frequently observed proteins. We have identified alterations in a number of specific UP in AR, primarily relating to MHC antigens, the complement cascade and extra-cellular matrix proteins. A subset of proteins (uromodulin, SERPINF1 and CD44), have been further cross-validated by ELISA in an independent set of urine samples, for significant differences in the abundance of these UP in AR.This label-free, semi-quantitative approach for sampling the urinary proteome in normal and disease states provides a robust and sensitive method for detection of UP for serial, non-invasive clinical monitoring for graft rejection after kidney transplantation.
View details for DOI 10.1002/prca.200900124
View details for Web of Science ID 000274264900004
View details for PubMedID 20543976
View details for PubMedCentralID PMC2883247
Large-Scale Multiplexed Quantitative Discovery Proteomics Enabled by the Use of an O-18-Labeled "Universal" Reference Sample
JOURNAL OF PROTEOME RESEARCH
2009; 8 (1): 290-299
The quantitative comparison of protein abundances across a large number of biological or patient samples represents an important proteomics challenge that needs to be addressed for proteomics discovery applications. Herein, we describe a strategy that incorporates a stable isotope (18)O-labeled "universal" reference sample as a comprehensive set of internal standards for analyzing large sample sets quantitatively. As a pooled sample, the (18)O-labeled "universal" reference sample is spiked into each individually processed unlabeled biological sample and the peptide/protein abundances are quantified based on (16)O/(18)O isotopic peptide pair abundance ratios that compare each unlabeled sample to the identical reference sample. This approach also allows for the direct application of label-free quantitation across the sample set simultaneously along with the labeling-approach (i.e., dual-quantitation) since each biological sample is unlabeled except for the labeled reference sample that is used as internal standards. The effectiveness of this approach for large-scale quantitative proteomics is demonstrated by its application to a set of 18 plasma samples from severe burn patients. When immunoaffinity depletion and cysteinyl-peptide enrichment-based fractionation with high resolution LC-MS measurements were combined, a total of 312 plasma proteins were confidently identified and quantified with a minimum of two unique peptides per protein. The isotope labeling data was directly compared with the label-free (16)O-MS intensity data extracted from the same data sets. The results showed that the (18)O reference-based labeling approach had significantly better quantitative precision compared to the label-free approach. The relative abundance differences determined by the two approaches also displayed strong correlation, illustrating the complementary nature of the two quantitative methods. The simplicity of including the (18)O-reference for accurate quantitation makes this strategy especially attractive when a large number of biological samples are involved in a study where label-free quantitation may be problematic, for example, due to issues associated with instrument platform robustness. The approach will also be useful for more effectively discovering subtle abundance changes in broad systems biology studies.
View details for DOI 10.1021/pr800467r
View details for Web of Science ID 000262171100030
View details for PubMedID 19053531
View details for PubMedCentralID PMC2752204
A transcriptome database for astrocytes, neurons, and oligodendrocytes: A new resource for understanding brain development and function
JOURNAL OF NEUROSCIENCE
2008; 28 (1): 264-278
Understanding the cell-cell interactions that control CNS development and function has long been limited by the lack of methods to cleanly separate neural cell types. Here we describe methods for the prospective isolation and purification of astrocytes, neurons, and oligodendrocytes from developing and mature mouse forebrain. We used FACS (fluorescent-activated cell sorting) to isolate astrocytes from transgenic mice that express enhanced green fluorescent protein (EGFP) under the control of an S100beta promoter. Using Affymetrix GeneChip Arrays, we then created a transcriptome database of the expression levels of >20,000 genes by gene profiling these three main CNS neural cell types at various postnatal ages between postnatal day 1 (P1) and P30. This database provides a detailed global characterization and comparison of the genes expressed by acutely isolated astrocytes, neurons, and oligodendrocytes. We found that Aldh1L1 is a highly specific antigenic marker for astrocytes with a substantially broader pattern of astrocyte expression than the traditional astrocyte marker GFAP. Astrocytes were enriched in specific metabolic and lipid synthetic pathways, as well as the draper/Megf10 and Mertk/integrin alpha(v)beta5 phagocytic pathways suggesting that astrocytes are professional phagocytes. Our findings call into question the concept of a "glial" cell class as the gene profiles of astrocytes and oligodendrocytes are as dissimilar to each other as they are to neurons. This transcriptome database of acutely isolated purified astrocytes, neurons, and oligodendrocytes provides a resource to the neuroscience community by providing improved cell-type-specific markers and for better understanding of neural development, function, and disease.
View details for DOI 10.1523/JNEUROSCI.4178-07.2008
View details for Web of Science ID 000252242900029
View details for PubMedID 18171944
High dynamic range characterization of the trauma patient plasma proteome
MOLECULAR & CELLULAR PROTEOMICS
2006; 5 (10): 1899-1913
Although human plasma represents an attractive sample for disease biomarker discovery, the extreme complexity and large dynamic range in protein concentrations present significant challenges for characterization, candidate biomarker discovery, and validation. Herein we describe a strategy that combines immunoaffinity subtraction and subsequent chemical fractionation based on cysteinyl peptide and N-glycopeptide captures with two-dimensional LC-MS/MS to increase the dynamic range of analysis for plasma. Application of this "divide-and-conquer" strategy to trauma patient plasma significantly improved the overall dynamic range of detection and resulted in confident identification of 22,267 unique peptides from four different peptide populations (cysteinyl peptides, non-cysteinyl peptides, N-glycopeptides, and non-glycopeptides) that covered 3,654 different proteins with 1,494 proteins identified by multiple peptides. Numerous low abundance proteins were identified, exemplified by 78 "classic" cytokines and cytokine receptors and by 136 human cell differentiation molecules. Additionally a total of 2,910 different N-glycopeptides that correspond to 662 N-glycoproteins and 1,553 N-glycosylation sites were identified. A panel of the proteins identified in this study is known to be involved in inflammation and immune responses. This study established an extensive reference protein database for trauma patients that provides a foundation for future high throughput quantitative plasma proteomic studies designed to elucidate the mechanisms that underlie systemic inflammatory responses.
View details for DOI 10.1074/mcp.M600068/MCP200
View details for Web of Science ID 000241519300017
View details for PubMedID 16684767
View details for PubMedCentralID PMC1783978