About

Finale Doshi-Velez is the Herchel Smith Professor of Computer Science at Harvard University, affiliated with the Harvard John A. Paulson School of Engineering and Applied Sciences. Her primary teaching area is Computer Science. Her research areas include applied mathematics, machine learning, artificial intelligence, computation and society, and computational and data science. She has been recognized for her contributions to mentoring, receiving the Everett Mendelsohn Excellence in Mentoring Award from the Graduate Student Council. Her work focuses on advancing AI and machine learning, shaping how these technologies influence our lives, and exploring their future development.

Selected publications

• Personalising AI Assistance Based on Overreliance Rate in AI-Assisted Decision Making

  2025-03-19 · 14 citations

  articleSenior author
  PublisherDOI

• Preference-based assistance optimization for lifting and lowering with a soft back exosuit

  Science Advances · 2025-04-09 · 6 citations

  articleOpen access

  Wearable robotic devices have become increasingly prevalent in both occupational and rehabilitative settings, yet their widespread adoption remains inhibited by usability barriers related to comfort, restriction, and noticeable functional benefits. Acknowledging the importance of user perception in this context, this study explores preference-based controller optimization for a back exosuit that assists lifting. Considering the high mental and metabolic effort discrete motor tasks impose, we used a forced-choice Bayesian Optimization approach that promotes sampling efficiency by leveraging domain knowledge about just noticeable differences between assistance settings. Optimizing over two control parameters, preferred settings were consistent within and uniquely different between participants. We discovered that overall, participants preferred asymmetric parameter configurations with more lifting than lowering assistance, and that preferences were sensitive to user anthropometrics. These findings highlight the potential of perceptually guided assistance optimization for wearable robotic devices, marking a step toward more pervasive adoption of these systems in the real world.

  PublisherDOI

• Federated ADMM from Bayesian Duality

  ArXiv.org · 2025-06-16

  preprintOpen access

  We propose a new Bayesian approach to generalize the federated Alternating Direction Method of Multipliers (ADMM). We show that the solutions of variational-Bayesian (VB) objectives are associated with a duality structure that not only resembles the structure of ADMM's fixed-points but also generalizes it. For example, ADMM-like updates are recovered when the VB objective is optimized over the isotropic-Gaussian family, and new non-trivial extensions are obtained for other exponential-family distributions. These extensions include a Newton-like variant that converges in one step on quadratic objectives and an Adam-like variant that yields up to 7% accuracy boosts for deep heterogeneous cases. Our work opens a new Bayesian way to generalize ADMM and other primal-dual methods.

  PublisherOA PDFDOI

• Brief mentalizing imagery therapy reduces stress and enhances positive psychological traits in family dementia caregivers: mediation by mindfulness

  Aging & Mental Health · 2025-06-24

  articleOpen access

  OBJECTIVES: Family dementia caregivers exhibit high rates of chronic psychological stress. We aimed to determine the impact of mentalizing imagery therapy (MIT), a mindfulness and guided imagery approach, on perceived stress and positive psychological traits. We further investigated the role of dispositional mindfulness on these effects. METHOD: = 46) of 4-week MIT versus a psychosocial support group for family dementia caregivers. Measures of perceived stress, resilience, and other positive psychological traits were administered at baseline, post-group, and four months. Longitudinal analyses were conducted with mixed linear models, and mediation analyses with bootstrapping, in R. RESULTS: MIT demonstrated statistically significant benefits relative to the support group for perceived stress, resilience, spiritual well-being, and other positive traits. Changes from baseline to post-group in mindfulness significantly mediated the relationships between group and most outcomes. CONCLUSION: In this pilot trial, MIT reduced perceived stress and improved positive psychological traits. Mediation by dispositional mindfulness was demonstrated. Future research in larger samples should be aimed at confirming the benefits of short-term MIT on reducing stress and improving positive psychological traits, and studying the time course of mediation effects by mindfulness.

  PublisherOA PDFDOI

• Connecting Federated ADMM to Bayes

  ArXiv.org · 2025-01-28

  preprintOpen accessSenior author

  We provide new connections between two distinct federated learning approaches based on (i) ADMM and (ii) Variational Bayes (VB), and propose new variants by combining their complementary strengths. Specifically, we show that the dual variables in ADMM naturally emerge through the 'site' parameters used in VB with isotropic Gaussian covariances. Using this, we derive two versions of ADMM from VB that use flexible covariances and functional regularisation, respectively. Through numerical experiments, we validate the improvements obtained in performance. The work shows connection between two fields that are believed to be fundamentally different and combines them to improve federated learning.

  PublisherOA PDFDOI

• Strategically Linked Decisions in Long-Term Planning and Reinforcement Learning

  ArXiv.org · 2025-05-22

  preprintOpen accessSenior author

  Long-term planning, as in reinforcement learning (RL), involves finding strategies: actions that collectively work toward a goal rather than individually optimizing their immediate outcomes. As part of a strategy, some actions are taken at the expense of short-term benefit to enable future actions with even greater returns. These actions are only advantageous if followed up by the actions they facilitate, consequently, they would not have been taken if those follow-ups were not available. In this paper, we quantify such dependencies between planned actions with strategic link scores: the drop in the likelihood of one decision under the constraint that a follow-up decision is no longer available. We demonstrate the utility of strategic link scores through three practical applications: (i) explaining black-box RL agents by identifying strategically linked pairs among decisions they make, (ii) improving the worst-case performance of decision support systems by distinguishing whether recommended actions can be adopted as standalone improvements or whether they are strategically linked hence requiring a commitment to a broader strategy to be effective, and (iii) characterizing the planning processes of non-RL agents purely through interventions aimed at measuring strategic link scores - as an example, we consider a realistic traffic simulator and analyze through road closures the effective planning horizon of the emergent routing behavior of many drivers.

  PublisherOA PDFDOI

• Estimating Upper Extremity Fugl-Meyer Assessment Scores From Reaching Motions Using Wearable Sensors

  IEEE Journal of Biomedical and Health Informatics · 2025-02-13 · 8 citations

  article

  The Fugl Meyer Assessment (FMA) is a widely-used assessment for tracking motor function recovery post-stroke. Due to the limited access to rehabilitation, there exists a need for remote and automated assessment solutions. Wearable sensors and data-driven methods have shown promise for enabling automatic upper extremity FMA (FMA-UE) estimation, but minimizing user input motion and aligning with current clinical activities will aid the adoption of sensor-based assessments. In this work, we present an FMA-UE estimator which can make score predictions for a key subset of the assessment (70$\% $ of all items) using data from inertial measurement units (IMUs) placed on the arms and the trunk from three volitional reaching motions representative of functional daily activities. We collected a dataset of eleven stroke participants performing a subset of FMA-UE, and three reaching motions. The FMA-UE of each participant was assessed by an occupational therapist providing the labeled score for the training data. The estimator was trained on windowed data during FMA-UE motions and was able to make score estimates from reaching motions. Through leave-one-subject-out cross validation, the estimator achieved a normalized RMSE of 7$\% $, which is comparable to or below the established minimal clinically important difference and minimal detectable change of FMA-UE of post-stroke individuals. Comparison experiments of various model designs also revealed the importance of trunk-based features inspired by compensation strategies common post stroke and features extracted from the hand sensor. The proposed estimator has the potential to broaden the possibility of automatic assessment via wearable sensors.

  PublisherDOI

• A Deployed Online Reinforcement Learning Algorithm in an Oral Health Clinical Trial

  Proceedings of the AAAI Conference on Artificial Intelligence · 2025-04-11 · 3 citations

  articleOpen access

  Dental disease is a prevalent chronic condition associated with substantial financial burden, personal suffering, and increased risk of systemic diseases. Despite widespread recommendations for twice-daily tooth brushing, adherence to recommended oral self-care behaviors remains sub-optimal due to factors such as forgetfulness and disengagement. To address this, we developed Oralytics, a mHealth intervention system designed to complement clinician-delivered preventative care for marginalized individuals at risk for dental disease. Oralytics incorporates an online reinforcement learning algorithm to determine optimal times to deliver intervention prompts that encourage oral self-care behaviors. We have deployed Oralytics in a registered clinical trial. The deployment required careful design to manage challenges specific to the clinical trials setting in the U.S. In this paper, we (1) highlight key design decisions of the RL algorithm that address these challenges and (2) conduct a re-sampling analysis to evaluate algorithm design decisions. A second phase (randomized control trial) of Oralytics is planned to start in spring 2025.

  PublisherOA PDFDOI

• Genomic Drivers of Coronary Artery Disease and Risk of Future Outcomes After Coronary Angiography

  JAMA Network Open · 2025-01-21 · 3 citations

  articleOpen access

  Importance: Disease characteristics of genetically mediated coronary artery disease (CAD) on coronary angiography and the association of genomic risk with outcomes after coronary angiography are not well understood. Objective: To assess the angiographic characteristics and risk of post-coronary angiography outcomes of patients with genomic drivers of CAD: familial hypercholesterolemia (FH), high polygenic risk score (PRS), and clonal hematopoiesis of indeterminate potential (CHIP). Design, Setting, and Participants: A retrospective cohort study of 3518 Mass General Brigham Biobank participants with genomic information who underwent coronary angiography was conducted between July 18, 2000, and August 1, 2023. Exposures: The presence of a genomic risk factor of CAD, defined as FH variant, high CAD PRS, or CHIP driver variation. Main Outcomes and Measures: Coronary artery disease presentation (stable or acute), angiographic CAD characteristics (severity and burden), angiographic outcomes (repeat angiogram, revascularization, and in-stent restenosis), and clinical outcomes (heart failure and all-cause mortality). Results: Among 3518 participants (2467 [70.1%] male; median age, 64.0 [IQR, 55.0-72.0] years), 1509 (42.9%) had at least 1 genomic driver of CAD (26 FH, 1191 high CAD PRS, and 466 CHIP) that was associated with the presentation of acute coronary syndromes (adjusted odds ratio, 2.67; 95% CI, 2.19-3.26) and with the presence, burden, and severity of angiographic CAD. This association was driven by FH and CAD PRS. One SD of CAD PRS was associated with a 12.51-point higher Gensini score. During 9 years of follow-up, there was an increased risk among FH carriers for a repeat angiogram (adjusted hazard ratio [AHR], 1.70; 95% CI, 1.02-2.83), and revascularization (AHR, 1.97; 95% CI, 1.02-3.80), and among people with high CAD PRS (repeat angiogram: AHR, 1.79; 95% CI, 1.45-2.22; revascularization: AHR, 1.85; 95% CI, 1.37-2.50; and in-stent restenosis: AHR, 3.89; 95% CI, 2.16-7.01). CHIP carriers had no significant increase in angiographic outcomes but were at higher risk of heart failure (AHR, 1.58; 95% CI, 1.04-2.40) and all-cause mortality (AHR, 1.78; 95% CI, 1.47-2.16). Conclusions and Relevance: The findings of this study suggest that germline monogenic and polygenic risk are associated with acute coronary syndromes presentation, severity and burden of atherosclerosis, and risk of repeat angiogram, revascularization, and in-stent restenosis. CHIP variant status is associated with incident heart failure and mortality after coronary angiography.

  PublisherOA PDFDOI

• Contrastive Explanations That Anticipate Human Misconceptions Can Improve Human Decision-Making Skills

  2025-04-24 · 20 citations

  article
  PublisherDOI

Awards & honors

• Everett Mendelsohn Excellence in Mentoring Award (2019)