JeevShastra

Continuum is a clinical chart for hospital wards where the monitors, lab analysers and referral letters are not connected to any system. In most Indian district and tertiary hospitals, staff read vitals off a bedside monitor and copy them by hand, lab reports arrive as paper or a photo on WhatsApp, and the ward round is spoken, not typed. Errors creep in at every copy, and the handover at shift change depends on memory. Continuum turns those inputs into a verified record. A nurse photographs the monitor, lab slip, or referral letter. The model reads the image and proposes each value, citing the exact region of the photo it came from. A crop-based verifier checks every citation against the pixels, and a plausibility check sets aside anything physiologically impossible, with the reason shown. The ward round is recorded as audio and transcribed with speaker turns, and the vitals spoken in the round are proposed the same way. Nothing enters the record until a clinician accepts it. Contradictions between sources (a referral letter saying no known allergies against a verified penicillin allergy) are flagged, and the clinician resolves them with a typed rationale that is stored permanently. The record is an append-only ledger in Postgres: every fact, decision and void is kept, so the audit trail is complete. From the verified facts, the model drafts an SBAR handover in which every sentence carries evidence chips linking back to its sources. Readiness checks block discharge until open conflicts and tasks are resolved. ICU transfers automatically switch the alert bands. I built this in ten days with Muse Code, and tested it the way I would use it on a ward. All data in the demo is synthetic.