Mohammad Asadi

Frontier multimodal models readily generate detailed image descriptions and elaborate reasoning traces (including pathology-biased clinical findings) for images they were never shown, a phenomenon we term "mirage reasoning." In the most extreme case, a small text-only model reached the top of a chest X-ray question-answering benchmark without access to any images. We introduce the Mirage Score, the Phantom-0 benchmark, and the B-Clean decontamination framework to measure and contain the effect.

MARCUS (Multimodal Autonomous Reasoning and Chat for Ultrasound and Signals) is an agentic vision-language assistant that reads and reasons over ECG, echocardiography, and cardiac MRI, individually or jointly. Modality-specific expert models are coordinated by a multimodal orchestrator that, by design, resists the "mirage reasoning" failure mode.

OpenMyHeartCounts (OpenMHC) is the largest and most comprehensive open-access wearable health dataset released to date, published alongside open-source implementations of recent wearable foundation models. Derived from over a decade of data collected through the My Heart Counts study app, it includes more than 60 million hours of wearable data across 19 sensor channels and 169 linked variables from 11,894 consenting participants. A unified open benchmark enables standardized comparison of wearable health models across downstream prediction, multivariate data imputation, and time-series forecasting.

NeuroQA is a large-scale benchmark for visual question answering in 3D brain MRI, with 56,953 question-answer pairs from 12,977 subjects across 12 datasets. It spans ages 5 to 104 and five clinical domains: Alzheimer’s disease, Parkinson’s disease, tumors, white matter disease, and neurodevelopment. Unlike prior medical VQA efforts that operate on 2D slices or rely on narrow diagnostic labels, NeuroQA pairs every item with a full 3D volume and evaluates 11 clinically grounded reasoning skills across yes/no, multiple-choice, and open-ended formats.