🤝 Top Collaborators
.png&w=640&q=75)
🤓 Latest Submissions
CareCoordinator
Patients and caregivers are often overwhelmed by scattered medical records, from discharge summaries and prescriptions to lab reports and clinical notes, making it hard to prepare for appointments or track a health history. General-purpose AI tools can introduce unsupported facts, while cloud services may expose sensitive health data to external APIs. CareCoordinator addresses both challenges through local inference and verification-first processing. It runs on an AMD Instinct MI300X GPU (192 GB) using ROCm and vLLM to serve Gemma-4-31B-IT across six specialized agents with continuous batching. When a user uploads documents and describes a goal, a Main Agent creates a structured Care Contract and delegates tasks to four parallel Worker Agents: Intake, Document Classifier, Record Summarizer, and Medication Checker. A Manager/Compiler Agent combines their findings into four sections: a plain-language summary, questions for the doctor, a medication checklist, and a clinician review section. The key differentiator is that verification is enforced through deterministic Python checks rather than left to the model. The backend validates inline citations against the original source text, while claims it cannot ground are marked “cannot confirm” and routed to clinician review instead of being presented as fact. A multi-layer safety gate halts normal processing when a configured emergency indicator is triggered, while historical risks are flagged for review. Users must approve a summary before it becomes final, and the “Improve” feature routes feedback to the relevant worker for revision and re-verification. In a demo with synthetic patient records, CareCoordinator generated a cited summary at a 7.5 reading grade level in 83 seconds, with 52 statements checked against the sources. It helps patients and caregivers organize their records and prepare for clinical conversations without diagnosing conditions or giving medical advice.
18 Oct 2026
AIRA
When someone calls the ER, they're rarely calm or organized. They're usually stressed, in pain, or trying to help someone who is. AIRA is a voice tool we built to handle that messy reality. It lets patients or caregivers simply explain what's going on in their own words. As they talk, AssemblyAI's Universal-3 Pro handles the real-time transcription. AIRA keeps track of what's already been said and only asks follow-up questions when something important is missing things like when the symptoms started, how severe they are, current medications, or allergies. The goal is to get the information staff need without forcing someone through a rigid checklist. The real bottleneck for ER staff isn't just taking the call. It's what happens after. Handing a nurse a massive raw transcript doesn't help much when they're already dealing with a busy emergency department. So, we use AssemblyAI's LeMUR to turn the conversation into a concise, structured triage brief. It appears on a staff dashboard, with every detail linked back to an exact timestamp. If a doctor needs to verify something, they can jump straight to that moment in the conversation instead of searching through an entire transcript. We also use AssemblyAI's Sentiment Analysis and Disfluency Detection to pick up on signs that a caller may be stressed, hesitant, or struggling to communicate. On the backend, JSON-Schema Tool Calling handles things like patient lookup and keeps the extracted information in consistent, structured fields. We were careful to make sure AIRA stays assistive. It doesn't diagnose the patient or make medical decisions. Staff review the generated brief, edit anything that needs correcting, request more information if necessary, and explicitly approve the case before it enters the clinical workflow.
30 Sep 2026
