Back to Work
Case StudyFull-Stack + Applied ML20261 min read

EmotiCare

A therapy intake app that reads a patient's emotional state from a few free-text sentences and hands the therapist context before the first session.

Health TechEmotion AIIntake Automation
1 / 3

Intake opens with a single prompt: patients describe how they've been feeling in their own words, up to 500 characters.

01Challenge

Therapy intake usually starts cold. Patients struggle to put feelings into words in a clinical setting, and therapists spend the first session just gathering basic emotional context. Urgent cases look the same as routine ones until someone reads every form by hand.

02Solution

EmotiCare moves that context up front. A patient describes how they've been feeling in plain text; a classifier sorts it into one of six primary emotions and shows its confidence and reasoning, which the patient can confirm or correct before booking. Therapists get a color-coded dashboard that surfaces each patient's own words, detected emotion, and urgency at a glance.

  • Free-text feeling entry classified into six primary emotions, each with a confidence score and short rationale
  • Human-in-the-loop confirmation: the patient accepts or adjusts the detected emotion before it's saved
  • Timezone-aware booking with three urgency tiers for triage
  • Therapist dashboard with color-coded emotion badges and urgency filtering
A machine's read on your feelings should be a starting point, not a verdict, so every detection is the patient's to confirm or correct.
03Results

An intake MVP that gives therapists a running start. Every booking arrives tagged with the patient's own words, a detected emotion, and an urgency level. A two-layer spam filter, combining client-side heuristics with model reasoning, keeps gibberish out of the queue.

04Tech Stack
Next.jsReactTypeScriptNeon PostgresGroq (Llama 3.3)

Want to see it in action?

Live Demo