nikam.dev ← All cases Telegram →
nikam.dev / Cases / Doctor ↔ patient platform
Live · NDA · Doctor + Patient + TG bot

A platform between a health expert and their patient.

The doctor hands the patient a treatment protocol - the patient lives it every day: ticks meds off in Telegram, logs how they feel, keeps lab results in one history. The doctor gets a ready, structured stage-based report instead of storytelling at the visit. The doctor console as a two-role platform is designed separately. Under NDA.

Type
Web SPA + TG bot
Audience
Doctor ↔ patient
Stack
FastAPI · SQLite · Aiogram
Status
Live · NDA
- The problem

The doctor hands out a protocol and loses contact with it until the next visit.

Between two visits the doctor has a blind zone. The patient is on their own at home: forgot something, mixed something up, didn't understand something. The doctor finds out a month later, on the next visit. Treatment runs «somehow», not «to plan». And the doctor has no instrument to see it coming.

Pain 01
The protocol is a sheet with 30 items
The patient gets a dense document with dozens of meds, different doses, timing windows («fasting / 30 min before meals / after meals / at bedtime»), courses of different lengths, rotations. Impossible to memorise. Two weeks in, half of it falls out.
Pain 02
Forgot - no one will remind
Missed the morning dose, didn't remember at work, by night it's too late. No timing reminders, no «5 days of course left» alerts, no feedback loop. The patient is alone.
Pain 03
Symptoms - in head, not in data
«Feels like I slept worse this week». «Gut seems calmer». Between two visits the doctor accumulates a stack of patient recollections, not data. Correlations with doses / course phases are invisible.
Pain 04
Lab results - scattered in email and WhatsApp
PDF results saved on the phone, paper ones filed at the clinic, others stuck in email. To see Hb / Ht trends across 2 years means collecting it by hand. The doctor starts from scratch each time.
Pain 05
Reporting back to the doctor - by storytelling
At the follow-up: «how was the month?» - «fine, I think». No systematic information. The doctor adjusts the protocol blind.

The protocol became a product the patient opens every morning.

Single source of truth - a structured protocol. From it, an engine builds: today's checklist by slot, course calendar, situational rules, shopping lists, doctor reports. Nothing for the patient to memorise.

«Today» - auto-checklist by slot
Fasting, 30 min before meals, after breakfast, between meals, after dinner, bedtime. The engine computes what's due today, accounting for courses, rotations, sanitation phases, delayed starts. One-tap checkboxes.
Telegram bot - checklists right in the chat
At 09:00 the patient gets a morning list with inline checkboxes in Telegram. At 14:00 a ping on the unchecked. At 21:00 a «possibly missed?» follow-up plus a quick symptom poll (sleep / energy / GI 1-5). Two-way sync: ticks in the bot show in Web and vice versa.
Symptoms in data, not in memory
Tap-log: sleep, energy, GI, swelling, blood pressure. Every day, one tap. A month in, correlations with course phases emerge. The doctor sees dynamics, not retellings.
Lab results in one history
All lab data in one system. Hb / Ht / erythrocytes / vitamins in line charts across the years. New result added through one screen, not into yet another folder.
Shopping list with direct links
What and where to buy, grouped by supplier (iHerb, Argo). Tagged «ahead-of-time» / «now», with «bought» checkboxes. No more «forgot to order - ran out».
PDF report for the doctor - by stages
«Save PDF» button. Output: a structured report for the period - adherence by protocol stage, average symptom scores, lab trends, biggest misses. The doctor sees a month at a glance in 30 seconds.
Situational rules
Constipation → med X. Skin flare → med Y. Pain → med Z. Cards surface when the patient logs the matching symptom. The patient handles minor decisions, the doctor isn't pinged for every small thing.

One engine that holds the whole protocol complexity.

The platform's main job: turn a dense medical document into «what, right now, to do». That needs structured parsing of the protocol plus an engine with a clean time model.

/ 01
Structured protocol
Source of truth - JSON marked up by fields: drug, dose, timing (relative-to-meals), frequency, length, start date, delayed starts, rotations. Parsed from the doctor's document, cross-checked by four independent AI models and verified by a human - medicine doesn't forgive mistakes. Updates monthly.
/ 02
Today engine
Handles: course windows, forward start dates, two-phase, every-other-day, 1mo-on/off, multi-dose slots, rotations, sanitation phases, delayed starts. Computes «what's due right now» with all of that in one pass. The patient sees a clean list, no logic.
/ 03
Sync between Web and Telegram
Tick in web → visible in bot in 12 seconds. Tick in bot → visible in web instantly. One source of truth - the adherence table. No «versions», no drift.
/ 04
Doctor reports
Per period: adherence by stage, symptom dynamics, lab trends, done / skipped. Output - a PDF the doctor reads in a minute.
- Outcome

What changes for the doctor and the patient.

0
Paperwork or memorisation on the patient. The protocol lives in the app and in the bot, not in the head. No more «forgot», no more «lost the document».
3 pings a day
Reminders on timing: morning checklist, midday ping on the unchecked, evening well-being poll. The patient isn't left alone between visits.
30 sec
The doctor's PDF report. Before: storytelling plus 20 minutes of digging. Now: a structured document for the period, everything on one page.

Similar task in your practice?

If you run patients through long, complex protocols (functional medicine, rehab, preventive, hormonal correction, chronic conditions) - let's talk about how this platform looks in your scenario. Adapting it to a specific specialty is a separate piece of work.

Message on Telegram → nick@nikam.dev