This dataset contains 221 published patent records filed between 2015 and 2026 that combine Medical AI claim language with clinical-analytics elements like training datasets, patient records or clinical decision logic. The count reflects a targeted search string rather than every AI-in-healthcare filing globally, so it should be read as a bounded sample of the field, not a total. Because publication lags filing by roughly 18 months, the most recent one to two years will likely be revised upward as more records surface.

The ranked list covers 100 companies, with the leading assignee holding 14 of the 221 records and the fifth-place assignee holding 7. No company controls a dominant share: the top 5 combined account for only 22.6% of all records and the top 10 for 33.0%. That means most of the filing activity sits with a long tail of universities, startups and individual inventors rather than a handful of large incumbents.

G06N (computing based on AI models) and G16H (healthcare informatics) are the two largest classes, covering 65.6% and 58.4% of the 221 records respectively. A61B (diagnosis and surgery), G06T (image data processing) and G06F/G06V follow as a secondary tier. Because a single record can carry several IPC classes, these percentages add up to more than 100% and describe overlapping claim coverage rather than a strict breakdown.

Yes, based on the years the dataset can treat as complete: filings grew from 10 in 2021 to 30 in 2024, a +200% increase over that three-year span, and 2025 recorded the highest raw count so far at 95 published records. The 2025 and 2026 figures are still incomplete due to normal publication lag, so they understate true filing activity for those years rather than signalling a slowdown.

Filing density is thin in a few identifiable branches relative to their clinical relevance, including closed-loop control of diagnostic AI systems (G05B overlaps with only 3.6% of records), federated model training directly on patient records, and real-time model inference on bedside or point-of-care devices. These are areas where claim space is comparatively open rather than crowded with prior art. A first-filer in one of these branches would face less direct claim overlap than in the densely claimed G06N and G16H classes.

A representative filing in this dataset, published as US20260188495A1 by University of Sharjah on 2026-07-02, claims an AI-based clinical decision support system that ingests brain scan data such as fMRI and EEG, analyses it using functional connectivity statistics with counterfactual explanations, and outputs a diagnosis plus attribution information identifying which connectivity features drove it. It illustrates how current claims combine multi-modal clinical input, model-based analysis and explainability into a single system rather than claiming any one element alone.