Nitin Gupta

MediQueue — Real-Time Clinic Queue System

Small clinics run queues on paper or by shouting names across the waiting room. Patients have no idea how long they'll wait, and receptionists repeat "you're next" dozens of times a day instead of focusing on actual patients. We built MediQueue to fix this with two synced screens: one for the receptionist to manage the queue, one for patients to see live status on their own phone — answering three questions: can a token be assigned in under 10 seconds, does the patient screen update without refreshing, and is the wait estimate based on real data instead of a guess? Process We started by mapping the real-time requirement first, since it was the riskiest part: Socket.IO broadcasting one shared "queue:updated" event to every connected screen, so receptionist and patient views never disagree. Data modeling came next — a Token schema with status (waiting/serving/served) and a singleton QueueState tracking the current token and a rolling average consultation time. The wait-time formula was the key design decision: instead of hardcoding "5 minutes per patient," we measured the actual duration of each consultation as it happened and folded it into a running average, so estimates get more accurate the longer the clinic uses it. We designed the UI in Stitch first — testing empty states, disabled buttons, and reconnect states before writing code — then built the backen Results Adding a patient and assigning a token takes one input field and the Enter key — well under 10 seconds. Every connected patient screen updates instantly when "Call Next" is clicked, with no manual refresh, verified across multiple simultaneous browser tabs. The wait-time estimate is no longer a static number — it's recalculated from real elapsed consultation time after every patient, so it reflects how this specific clinic actually runs rather than a generic assumption. Reflection I'd add basic authentication before deploying anywhere beyond a demo, since right now anyone with the link could act as the receptionist. I'd also build a small analytics view on top of the consultation-time data we're already collecting, since that history is currently calculated but never shown to the clinic owner. Lastly, I'd add a lightweight SMS/WhatsApp fallback for patients who'd rather not stare at a screen while waiting, rather than assuming everyone wants a live display.