Kenneth W. Mahaffey
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Research Subjects
Research Interests
Kenneth W. Mahaffey has a total of 24 co-authors. Here are the top co-authors by number of shared publications:
- Adeera Levin
Division of Nephrology, University of British Columbia, Vancouver, British Colombia, Canada
1 shared publication - Alejandro Lozano
Stanford University
1 shared publication - Brendon Lange Neuen
The George Institute for Global Health, University of New South Wales, Sydney, Australia
1 shared publication - Briggs Morrison
Syndax, 211 East 43rd Street, #900, New York, NY, 10017, USA
1 shared publication - Bruce Neal
The George Institute for Global Health, University of New South Wales, Sydney, Australia
1 shared publication - Christopher P. Cannon
Cardiovascular Division Brigham and Women's Hospital, Harvard Medical School Boston Massachusetts USA
1 shared publication
Latest Publications
- Open Access
Zero-Shot Clinical Trial Patient Matching with LLMs
Matching patients to clinical trials is a key unsolved challenge in bringing new drugs to market. Today, identifying patients who meet a trial's eligibility criteria is highly manual, taking up to 1 hour per patient. Automated screening is challenging, however, as it requires understanding unstructured clinical text. Large language models (LLMs) offer a promising solution. In this work, we explore their application to trial matching. First, we design an LLM-based system which, given a patient's medical history as unstructured clinical text, evaluates whether that patient meets a set of inclusion criteria (also specified as free text). Our zero-shot system achieves state-of-the-art scores on the n2c2 2018 cohort selection benchmark. Second, we improve the data and cost efficiency of our method by identifying a prompting strategy which matches patients an order of magnitude faster and more cheaply than the status quo, and develop a two-stage retrieval pipeline that reduces the number of tokens processed by up to a third while retaining high performance. Third, we evaluate the interpretability of our system by having clinicians evaluate the natural language justifications generated by the LLM for each eligibility decision, and show that it can output coherent explanations for 97% of its correct decisions and 75% of its incorrect ones. Our results establish the feasibility of using LLMs to accelerate clinical trial operations.
Open AccessZero-Shot Clinical Trial Patient Matching with LLMs
Matching patients to clinical trials is a key unsolved challenge in bringing new drugs to market. Today, identifying patients who meet a trial's eligibility criteria is highly manual, taking up to 1 hour per patient. Automated screening is challenging, however, as it requires understanding unstructured clinical text. Large language models (LLMs) offer a promising solution. In this work, we explore their application to trial matching. First, we design an LLM-based system which, given a patient's medical history as unstructured clinical text, evaluates whether that patient meets a set of inclusion criteria (also specified as free text). Our zero-shot system achieves state-of-the-art scores on the n2c2 2018 cohort selection benchmark. Second, we improve the data and cost efficiency of our method by identifying a prompting strategy which matches patients an order of magnitude faster and more cheaply than the status quo, and develop a two-stage retrieval pipeline that reduces the number of tokens processed by up to a third while retaining high performance. Third, we evaluate the interpretability of our system by having clinicians evaluate the natural language justifications generated by the LLM for each eligibility decision, and show that it can output coherent explanations for 97% of its correct decisions and 75% of its incorrect ones. Our results establish the feasibility of using LLMs to accelerate clinical trial operations.
- Open Access
- 1 May 2021
- Published by Wolters Kluwer Health in Stroke
- Vol. 52 (5) , 1545-1556
Effect of SGLT2 Inhibitors on Stroke and Atrial Fibrillation in Diabetic Kidney Disease
Background and Purpose: Chronic kidney disease with reduced estimated glomerular filtration rate or elevated albuminuria increases risk for ischemic and hemorrhagic stroke. This study assessed the effects of sodium glucose cotransporter 2 inhibitors (SGLT2i) on stroke and atrial fibrillation/flutter (AF/AFL) from CREDENCE (Canagliflozin and Renal Events in Diabetes With Established Nephropathy Clinical Evaluation) and a meta-analysis of large cardiovascular outcome trials (CVOTs) of SGLT2i in type 2 diabetes mellitus. Methods: CREDENCE randomized 4401 participants with type 2 diabetes mellitus and chronic kidney disease to canagliflozin or placebo. Post hoc, we estimated effects on fatal or nonfatal stroke, stroke subtypes, and intermediate markers of stroke risk including AF/AFL. Stroke and AF/AFL data from 3 other completed large CVOTs and CREDENCE were pooled using random-effects meta-analysis. Results: In CREDENCE, 142 participants experienced a stroke during follow-up (10.9/1000 patient-years with canagliflozin, 14.2/1000 patient-years with placebo; hazard ratio [HR], 0.77 [95% CI, 0.55–1.08]). Effects by stroke subtypes were: ischemic (HR, 0.88 [95% CI, 0.61–1.28]; n=111), hemorrhagic (HR, 0.50 [95% CI, 0.19–1.32]; n=18), and undetermined (HR, 0.54 [95% CI, 0.20–1.46]; n=17). There was no clear effect on AF/AFL (HR, 0.76 [95% CI, 0.53–1.10]; n=115). The overall effects in the 4 CVOTs combined were: total stroke (HR pooled , 0.96 [95% CI, 0.82–1.12]), ischemic stroke (HR pooled , 1.01 [95% CI, 0.89–1.14]), hemorrhagic stroke (HR pooled , 0.50 [95% CI, 0.30–0.83]), undetermined stroke (HR pooled , 0.86 [95% CI, 0.49–1.51]), and AF/AFL (HR pooled , 0.81 [95% CI, 0.71–0.93]). There was evidence that SGLT2i effects on total stroke varied by baseline estimated glomerular filtration rate ( P =0.01), with protection in the lowest estimated glomerular filtration rate (2 ]) subgroup (HR pooled , 0.50 [95% CI, 0.31–0.79]). Conclusions: Although we found no clear effect of SGLT2i on total stroke in CREDENCE or across trials combined, there was some evidence of benefit in preventing hemorrhagic stroke and AF/AFL, as well as total stroke for those with lowest estimated glomerular filtration rate. Future research should focus on confirming these data and exploring potential mechanisms. Registration: URL: https://www.clinicaltrials.gov ; Unique identifier: NCT02065791.
- 1 May 2021
- Published by Wolters Kluwer Health in Stroke
- Vol. 52 (5) , 1545-1556
Open AccessEffect of SGLT2 Inhibitors on Stroke and Atrial Fibrillation in Diabetic Kidney Disease
Background and Purpose: Chronic kidney disease with reduced estimated glomerular filtration rate or elevated albuminuria increases risk for ischemic and hemorrhagic stroke. This study assessed the effects of sodium glucose cotransporter 2 inhibitors (SGLT2i) on stroke and atrial fibrillation/flutter (AF/AFL) from CREDENCE (Canagliflozin and Renal Events in Diabetes With Established Nephropathy Clinical Evaluation) and a meta-analysis of large cardiovascular outcome trials (CVOTs) of SGLT2i in type 2 diabetes mellitus. Methods: CREDENCE randomized 4401 participants with type 2 diabetes mellitus and chronic kidney disease to canagliflozin or placebo. Post hoc, we estimated effects on fatal or nonfatal stroke, stroke subtypes, and intermediate markers of stroke risk including AF/AFL. Stroke and AF/AFL data from 3 other completed large CVOTs and CREDENCE were pooled using random-effects meta-analysis. Results: In CREDENCE, 142 participants experienced a stroke during follow-up (10.9/1000 patient-years with canagliflozin, 14.2/1000 patient-years with placebo; hazard ratio [HR], 0.77 [95% CI, 0.55–1.08]). Effects by stroke subtypes were: ischemic (HR, 0.88 [95% CI, 0.61–1.28]; n=111), hemorrhagic (HR, 0.50 [95% CI, 0.19–1.32]; n=18), and undetermined (HR, 0.54 [95% CI, 0.20–1.46]; n=17). There was no clear effect on AF/AFL (HR, 0.76 [95% CI, 0.53–1.10]; n=115). The overall effects in the 4 CVOTs combined were: total stroke (HR pooled , 0.96 [95% CI, 0.82–1.12]), ischemic stroke (HR pooled , 1.01 [95% CI, 0.89–1.14]), hemorrhagic stroke (HR pooled , 0.50 [95% CI, 0.30–0.83]), undetermined stroke (HR pooled , 0.86 [95% CI, 0.49–1.51]), and AF/AFL (HR pooled , 0.81 [95% CI, 0.71–0.93]). There was evidence that SGLT2i effects on total stroke varied by baseline estimated glomerular filtration rate ( P =0.01), with protection in the lowest estimated glomerular filtration rate (2 ]) subgroup (HR pooled , 0.50 [95% CI, 0.31–0.79]). Conclusions: Although we found no clear effect of SGLT2i on total stroke in CREDENCE or across trials combined, there was some evidence of benefit in preventing hemorrhagic stroke and AF/AFL, as well as total stroke for those with lowest estimated glomerular filtration rate. Future research should focus on confirming these data and exploring potential mechanisms. Registration: URL: https://www.clinicaltrials.gov ; Unique identifier: NCT02065791.
- Open Access
- 1 November 2020
- Published by Oxford University Press (OUP) in European Journal of Preventive Cardiology
- Vol. 27 (17) , 1887-1889
Diabetes and heart failure post-acute myocardial infarction: Important associations and need for evidence-based interventions
- 1 November 2020
- Published by Oxford University Press (OUP) in European Journal of Preventive Cardiology
- Vol. 27 (17) , 1887-1889
Open AccessDiabetes and heart failure post-acute myocardial infarction: Important associations and need for evidence-based interventions
- Open Access
- 4 December 2018
- Published by Springer Nature in Journal of Translational Medicine
- Vol. 16 (1) , 1-15
Reviewing the role of healthy volunteer studies in drug development
With the exception of genotoxic oncology drugs, first-in-human, Phase 1 clinical studies of investigational drugs have traditionally been conducted in healthy volunteers (HVs). The primary goal of these studies is to investigate the pharmacokinetics and pharmacodynamics of a novel drug candidate, determine appropriate dosing, and document safety and tolerability. When tailored to specific study objectives, HV studies are beneficial to manufacturers and patients alike and can be applied to both non-oncology and oncology drug development. Enrollment of HVs not only increases study accrual rates for dose-escalation studies but also alleviates the ethical concern of enrolling patients with disease in a short-term study at subtherapeutic doses when other studies (e.g. Phase 2 or Phase 3 studies) may be more appropriate for the patient. The use of HVs in non-oncology Phase 1 clinical trials is relatively safe but nonetheless poses ethical challenges because of the potential risks to which HVs are exposed. In general, most adverse events associated with non-oncology drugs are mild in severity, and serious adverse events are rare, but examples of severe toxicity have been reported. The use of HVs in the clinical development of oncology drugs is more limited but is nonetheless useful for evaluating clinical pharmacology and establishing an appropriate starting dose for studies in cancer patients. During the development of oncology drugs, clinical pharmacology studies in HVs have been used to assess pharmacokinetics, drug metabolism, food effects, potential drug–drug interactions, effects of hepatic and renal impairment, and other pharmacologic parameters vital for clinical decision-making in oncology. Studies in HVs are also being used to evaluate biosimilars versus established anticancer biologic agents. A thorough assessment of toxicity and pharmacology throughout the drug development process is critical to ensure the safety of HVs. With the appropriate safeguards, HVs will continue to play an important role in future drug development.
- 4 December 2018
- Published by Springer Nature in Journal of Translational Medicine
- Vol. 16 (1) , 1-15
Open AccessReviewing the role of healthy volunteer studies in drug development
With the exception of genotoxic oncology drugs, first-in-human, Phase 1 clinical studies of investigational drugs have traditionally been conducted in healthy volunteers (HVs). The primary goal of these studies is to investigate the pharmacokinetics and pharmacodynamics of a novel drug candidate, determine appropriate dosing, and document safety and tolerability. When tailored to specific study objectives, HV studies are beneficial to manufacturers and patients alike and can be applied to both non-oncology and oncology drug development. Enrollment of HVs not only increases study accrual rates for dose-escalation studies but also alleviates the ethical concern of enrolling patients with disease in a short-term study at subtherapeutic doses when other studies (e.g. Phase 2 or Phase 3 studies) may be more appropriate for the patient. The use of HVs in non-oncology Phase 1 clinical trials is relatively safe but nonetheless poses ethical challenges because of the potential risks to which HVs are exposed. In general, most adverse events associated with non-oncology drugs are mild in severity, and serious adverse events are rare, but examples of severe toxicity have been reported. The use of HVs in the clinical development of oncology drugs is more limited but is nonetheless useful for evaluating clinical pharmacology and establishing an appropriate starting dose for studies in cancer patients. During the development of oncology drugs, clinical pharmacology studies in HVs have been used to assess pharmacokinetics, drug metabolism, food effects, potential drug–drug interactions, effects of hepatic and renal impairment, and other pharmacologic parameters vital for clinical decision-making in oncology. Studies in HVs are also being used to evaluate biosimilars versus established anticancer biologic agents. A thorough assessment of toxicity and pharmacology throughout the drug development process is critical to ensure the safety of HVs. With the appropriate safeguards, HVs will continue to play an important role in future drug development.