About

Robert Harrington is the Arthur L. Bloomfield Professor of Medicine and also holds a courtesy appointment as a Professor of Health Policy at Stanford University. He is associated with the Center for Artificial Intelligence in Medicine & Imaging (AIMI) at Stanford, where his work focuses on the integration of artificial intelligence into medical and imaging sciences. His role involves advancing research in these areas, contributing to the development of innovative healthcare solutions through AI, and engaging in academic leadership within the center.

Research topics

• Medicine
• Internal medicine
• Computer Science
• Artificial Intelligence
• Intensive care medicine
• Cardiology
• Virology
• Machine Learning
• Pathology
• Environmental health
• Algorithm
• Emergency medicine
• Endocrinology

Selected publications

• APOLIPOPROTEIN A-I INFUSIONS AND CARDIOVASCULAR OUTCOMES IN PATIENTS WITH ACUTE MYOCARDIAL INFARCTION AND MODERATE RENAL IMPAIRMENT: INSIGHTS FROM THE AEGIS-II TRIAL

  Journal of the American College of Cardiology · 2025-03-29

  articleOpen accessSenior author
  PublisherDOI

• Relation of Low-Density Lipoprotein Cholesterol, High-Sensitivity C-Reactive Protein, and Lipoprotein(a) Each to Future Cardiovascular Events and Death After Acute Coronary Syndrome on High-Intensity Statin Therapy: An Analysis of the Placebo Arm of ODYSSEY OUTCOMES

  Circulation · 2025-04-07 · 21 citations

  letterOpen access

  Contains fulltext : 320894.pdf (Publisher’s version ) (Open Access)

  PublisherOA PDFDOI

• Multiplex Apolipoprotein Panel Improves Cardiovascular Event Prediction and Cardiovascular Outcome by Identifying Patients Who Benefit From Targeted PCSK9 Inhibitor Therapy

  Arteriosclerosis Thrombosis and Vascular Biology · 2025-09-25 · 3 citations

  articleOpen access

  BACKGROUND: Residual cardiovascular risk remains, despite achieving low-density lipoprotein cholesterol targets with high-intensity statins. Traditional risk scores are suboptimal. This study evaluated the prognostic utility of a 9-plex apolipoprotein panel in recent patients with acute coronary syndrome on statins and its role in predicting treatment benefit by alirocumab, a PCSK9 (proprotein convertase subtilisin/kexin type 9) inhibitor, enabling precision medicine. METHODS: Baseline serum samples from 11 843 participants in the ODYSSEY OUTCOMES trial ( https://www.clinicaltrials.gov ; Unique identifier: NCT01663402) were analyzed using mass spectrometry to measure Apo(a), ApoA-I, ApoA-II, ApoA-IV, ApoB, ApoC-I, ApoC-II, ApoC-III, and ApoE. Using logistic regression, probabilities of major adverse cardiovascular events (MACE) and all-cause death over a median follow-up of 2.9 years were estimated based on baseline apolipoproteins and lipid concentrations. Clinical performance was assessed by comparing the area under the curve (AUC) of 3 models: the apolipoprotein panel, the lipid panel (total cholesterol, high-density lipoprotein cholesterol, and triglycerides), and a combination. In addition, prediction models estimating the treatment benefit of alirocumab by the apolipoprotein panel were developed. RESULTS: The prognostic performance of the apolipoprotein panel for MACE showed an AUC (95% CI) of 0.648 (0.626–0.670), compared with 0.579 (0.557–0.602) for the lipid panel. For all-cause death, the apolipoprotein panel had an AUC of 0.699 (0.664–0.733), while the lipid panel had an AUC of 0.599 (0.564–0.635). Adding the apolipoprotein panel significantly improved the performance of the conventional lipid panel ( P <0.0001): AUC, 0.659 (0.637–0.681) for MACE and 0.724 (0.691–0.756) for all-cause death. Higher risk for MACE based on the baseline apolipoprotein panel was found to predict greater treatment benefit with alirocumab. CONCLUSIONS: A multiplex apolipoprotein panel led to better prediction of MACE and all-cause death, beyond lipids, in patients with postacute coronary syndrome on optimized statin therapy. The panel also predicts the treatment benefit of alirocumab. Further validation of this approach is now needed, and if confirmed and improved, it could lead to better disease prediction and management in the future.

  PublisherDOI

• Risk of Incident Diabetes Related to Lipoprotein(a), LDL Cholesterol, and Their Changes with Alirocumab: Post-Hoc Analyses of the ODYSSEY OUTCOMES Randomized Trial

  2025-02-06

  preprint

  <p dir="ltr"><a href="" target="_blank"><b>OBJECTIVE</b></a></p><p dir="ltr">Previous genetic and clinical analyses have associated lower lipoprotein(a) and LDL cholesterol with greater risk of new-onset type 2 diabetes (NOD). However, PCSK9 inhibitors such as alirocumab lower both lipoprotein(a) and LDL cholesterol without effect on NOD.</p><p dir="ltr"><b>RESEARCH DESIGN AND METHODS</b></p><p dir="ltr">In a post-hoc analysis of the ODYSSEY OUTCOMES trial (NCT01663402), we examined the joint prediction of NOD by baseline lipoprotein(a), LDL cholesterol, and insulin (or HOMA-IR) and their changes with alirocumab treatment. Analyses included 8107 patients with recent acute coronary syndrome on optimized statin therapy, without diabetes at baseline, assigned to alirocumab or placebo with median follow-up 2.4 years. Splines were estimated from logistic regression models.</p><p dir="ltr"><b>RESULTS</b></p><p dir="ltr">Lower baseline lipoprotein(a) and higher baseline insulin or HOMA-IR independently predicted 782 cases of NOD; baseline LDL cholesterol did not predict NOD. Alirocumab reduced lipoprotein(a) and LDL cholesterol without affecting insulin or NOD risk (odds ratio versus placebo [OR] 0.998; 95% CI 0.860-1.158). However, in logistic regression, decreased lipoprotein(a) and LDL cholesterol on alirocumab were independent, opposite predictors of NOD. <a href="" target="_blank">OR for NOD for 25% and 50% lipoprotein(a) reductions were 1.12 (95% CI 1.01</a>-1.23) and 1.24 (1.02-1.52). OR for NOD for 25% and 50% LDL cholesterol reductions were 0.88 (95% CI 0.80-0.97) and 0.77 (0.64-0.94).</p><p dir="ltr"><a href="" target="_blank"><b>CONCLUSIONS</b></a></p><p dir="ltr">Baseline lipoprotein(a) was inversely associated with risk of NOD. Alirocumab-induced reductions of lipoprotein(a) and LDL cholesterol were associated with increased and decreased risk of NOD, respectively, without net effect on NOD. Ongoing trials will determine the impact of larger and longer lipoprotein(a) reductions on NOD.</p>

  PublisherDOI

• Learning health system strategies in the AI era

  npj Health Systems · 2025-06-17 · 9 citations

  articleOpen access

  Abstract The learning health system (LHS) offers a framework to accelerate evidence generation and care improvement, yet widespread adoption remains limited. In this perspective, we explore strategies to operationalize the LHS in the era of artificial intelligence, including biomedical informatics and health information technology integration, workforce development, quality improvement, and data governance. We highlight promising institutional models and propose policy, educational, and financial reforms to support scalable, value-driven innovation in increasingly complex and resource-constrained health systems.

  PublisherOA PDFDOI

• Oral factor xia inhibitor milvexian after a recent acute coronary syndrome: Rationale and design of the phase 3 (Librexia ACS)

  American Heart Journal · 2025-02-20 · 12 citations

  articleOpen access

  BACKGROUND: Despite current antiplatelet therapy, patients remain at risk of recurrent ischemic events after acute coronary syndromes (ACS), which may reflect persistently elevated thrombin generation. Factor XIa inhibition reduces thrombin generation and may improve clinical outcomes with minimal bleeding risk. DESIGN: Librexia ACS (ClinicalTrials.gov NCT05754957) is a Phase 3, randomized, double-blind, placebo-controlled, event-driven trial to test the efficacy and safety of milvexian, an oral, selective factor XIa inhibitor, in addition to conventional antiplatelet therapy after a recent ACS. Eligibility criteria include symptoms of spontaneous ischemia, a diagnosis of ACS and cardiac biomarker elevation indicative of myonecrosis within 7 days before randomization, along with at least 2 risk-enhancing factors. Participants are randomly assigned to oral milvexian (25 mg twice daily) or a matched placebo. Randomization is stratified according to the planned duration and type of antiplatelet therapy. The primary efficacy endpoint is the time to first occurrence of the composite of cardiovascular death, myocardial infarction (MI), or ischemic stroke that will enroll approximately 16,000 patients with follow-up until 875 events are accrued. The first major secondary endpoint is time to the first occurrence of cardiovascular death, MI, ischemic stroke, major adverse limb events, and symptomatic venous thromboembolism. The principal safety endpoint is Bleeding Academic Research Consortium 3c or 5 bleeding. SUMMARY: The Librexia-ACS trial will determine the efficacy and safety of milvexian after ACS and will be the first trial to test whether factor XIa inhibition in addition to standard-of-care antiplatelet therapy reduces major adverse cardiovascular events without an increased risk of significant bleeding.

  PublisherDOI

• Impact of Apolipoprotein A-I Infusions on Cardiovascular Events Post-MI by Neutrophil-Lymphocyte Ratio and LDL-Cholesterol Levels

  JACC Advances · 2025-04-01 · 1 citations

  articleOpen access

  BACKGROUND: The AEGIS-II (ApoA-I Event Reducing in Ischemic Syndromes-II; NCT03473223) trial evaluated CSL112, a human plasma-derived apolipoprotein A-I therapy, for reducing cardiovascular events after acute myocardial infarction (AMI). Given CSL112's potential anti-inflammatory properties, we conducted an exploratory post hoc analysis to determine if its efficacy is influenced by baseline neutrophil-lymphocyte ratio (NLR), a marker of systemic inflammation, and low-density lipoprotein cholesterol (LDL-C). OBJECTIVES: The purpose of this study was to investigate the association of baseline NLR and cardiovascular events and explore whether NLR and LDL-C modify CSL112's efficacy in post-AMI patients. METHODS: A total of 18,219 participants with AMI, multivessel coronary artery disease, and additional cardiovascular risk factors were randomized to 4 weekly infusions of 6 g CSL112 or placebo. The primary endpoint was a composite of cardiovascular death, myocardial infarction, or stroke (major adverse cardiovascular events [MACE]). Cox proportional hazards models evaluated risk by dichotomized baseline NLR (>median vs ≤median). Treatment interactions with NLR and LDL-C (≥100 vs <100 mg/dL) were assessed. RESULTS: = 0.029). CONCLUSIONS: Baseline elevated NLR predicts MACE in post-AMI patients, and CSL112 showed an associated reduction in MACE in patients with elevated NLR and LDL-C ≥100 mg/dL.

  PublisherDOI

• Abstract 4360415: Small, Dense LDL-C and Conventional LDL-C Similarly Predict Cardiovascular Risk and Benefit of Alirocumab in Statin-Treated Patients With Recent Acute Coronary Syndrome

  Circulation · 2025-11-03

  article

  Introduction: Small, dense low-density lipoprotein (sdLDL) particles are believed to be a highly atherogenic subfraction of LDL due to prolonged residence time in circulation, greater adherence to and penetration of vascular endothelium, and higher susceptibility to oxidation. Automated biochemical measurement of sdLDL cholesterol (sdLDL-C) has demonstrated good fidelity to gold standard gradient ultracentrifugation or NMR spectroscopy. We evaluated the relationship of sdLDL-C and conventional LDL-C to risk of major adverse cardiovascular events (MACE) and treatment benefit of alirocumab in patients with recent acute coronary syndrome (ACS) receiving high-intensity or maximum-tolerated statin treatment. Methods: The analysis included 11,837 participants in the ODYSSEY OUTCOMES trial (NCT01663402) with recent ACS and LDL-C ≥70 mg/dL despite optimized statin treatment. At baseline prior to randomized treatment with the PCSK9 monoclonal antibody alirocumab (N=5917) or placebo (N=5920), sdLDL-C was measured using the Denka (Nigata, Japan) method on a Roche cobas autoanalyzer and LDL-C was calculated with the Friedewald formula. In the placebo group, natural cubic splines depicted the relationships of sdLDL-C, LDL-C, and their ratio to the risk of MACE (CV death, non-fatal myocardial infarction or ischemic stroke, hospitalization for unstable angina, and ischemia-driven coronary revascularization) and treatment hazard ratio (HR: alirocumab/placebo) as a function of sdLDL-C and LDL-C. Results: In Figure Panel A, the risk of MACE in the placebo group increased with concentrations of baseline sdLDL-C and LDL-C, with nearly superimposable splines. In Panel B, the relationship of sdLDL-C/LDL-C to risk of MACE in the placebo group (adjusted for LDL-C) showed no evidence of greater risk with greater sdLDL-C fraction. Overall, alirocumab reduced the risk of MACE (HR 0.87, 95% CI 0.79, 0.95). Panel C shows that the treatment HR did not vary significantly across the range of either LDL-C or sdLDL-C. Conclusion: In patients with recent ACS and LDL-C ≥70 mg/dL on optimized statin treatment, sdLDL-C and conventional LDL-C similarly predict risk of MACE and benefit of treatment with alirocumab. Measurement of sdLDL-C does not appear to provide additional prognostic or predictive information.

  PublisherDOI

• HIGH BLEEDING RISK AND CARDIOVASCULAR OUTCOMES: SECONDARY ANALYSIS OF THE ISCHEMIA TRIALS

  Journal of the American College of Cardiology · 2025-03-29

  articleOpen access
  PublisherDOI

• APOLIPOPROTEIN A-I INFUSIONS AND CARDIOVASCULAR OUTCOMES IN ACUTE MYOCARDIAL INFARCTION ACCORDING TO NEUTROPHIL-LYMPHOCYTE RATIO AND BASELINE LDLCHOLESTEROL LEVELS: INSIGHTS FROM THE AEGIS-II TRIAL

  Journal of the American College of Cardiology · 2025-03-29

  article
  PublisherDOI

Recent grants

• The ISCHEMIA Trial - SDCC

  NIH · $17.8M · 2011–2022

Frequent coauthors

• Philippe Gabríel Steg

  Université Paris Cité

  1919 shared

• Kenneth W. Mahaffey

  Center for Clinical Research (United States)

  1821 shared

• Harvey D. White

  Greenlane Clinical Centre

  1801 shared

• Deepak L. Bhatt

  Cornell University

  1378 shared

• Paul W. Armstrong

  Canadian VIGOUR Centre

  1113 shared

• Neal S. Kleiman

  Methodist Hospital

  998 shared

• Eric J. Topol

  Scripps Clinic

  966 shared

• Gregg W. Stone

  Icahn School of Medicine at Mount Sinai

  959 shared