AI-Powered Dynamic Mental Health Monitoring and Distress Prediction System for Victims of Atrocities

The receipts 🧾

Every red flag on this statement, in full. These are the four places it bites.

1. Exhibit A

   There is no longitudinal outcome data for this population and no ethical route to collecting it, so the distress prediction model has nothing to learn from and no way to be shown correct

2. It gets worse

   Continuous automated psychological monitoring of identified victims by a government platform, with alerts to district authorities, is a surveillance system as much as a support one, and the statement does not address consent, opt-out or who can see the score

3. Still reading?

   A predicted crisis that does not occur and an unpredicted one that does are both invisible in a demo, so the central claim cannot be evidenced at any point in the project

4. And the finisher

   The three-tier dashboard means individual victims' inferred mental states become visible up to national level, which is a design decision worth challenging rather than implementing

The damage report 📉

Every score this statement earned, and what each one actually costs you.

• Feasibility

  1/5

  You have picked a fight with physics, procurement, or both. One of them always wins.

  Predicting escalation of psychological distress requires longitudinal outcome data on this exact population and it does not exist and could not ethically be assembled, so the prediction model at the centre of the statement has neither training data nor any way to be validated.

• Innovation scope

  2/5

  Nothing here is new. Your only edge is execution — and execution is also everyone else's only edge.

  The interaction channels, the analysis techniques, the score, the alerting thresholds, the interventions to recommend and the three dashboard tiers are all enumerated, so the system is specified rather than the problem.

• Clarity

  4/5

  The ask is unambiguous, which quietly removes your favourite excuse.

  Detailed about what the system should produce and who it should alert, and it names the priority cases explicitly, but it never states what a distress score would be measured against or what would demonstrate that a predicted crisis was correctly predicted.

• Acceptance potential

  1/5

  The numbers do not like you. Bring something the numbers cannot see.

  This proposes ongoing automated psychological monitoring of rape and murder victims by a government system, built on prediction models that have no training data and no validation path — the surveillance concern and the technical hollowness compound each other, and neither is addressed in the statement.

• Effort

  Massive

  A semester of work wearing a hackathon costume. Something is getting cut; decide what now, not in week five.

  Periodic outreach across chatbot, IVR, SMS, mobile and web, longitudinal analysis, a prediction layer, threshold alerting, intervention recommendation and dashboards at district, state and national level is a large multi-channel system with heavy compliance requirements attached.

• Demo-ability

  Hard

  Near impossible to show working in five minutes, which is roughly five minutes more than you get.

  Longitudinal distress trajectories cannot be demonstrated in a hackathon window even in principle, and the population whose behaviour would validate the model is one you cannot approach — so any demonstration runs on invented case histories.

• Data

  None supplied

  No dataset comes with this one, so every accuracy figure you quote is a number about labels you invented.

  Nothing is provided with the statement. You are sourcing, cleaning and labelling it yourself, and that work is invisible in the demo but very visible in the questions.

The demo they will have already seen 🪞

Somewhere around 220–500 teams are heading here, and the description is doing the choosing for most of them. They will read the same brief, reach the same architecture, and build a version of the same demo you are planning. Being correct is the floor. If your five minutes could be swapped with the team before you and nobody in the room would notice, you have not picked badly — you have built predictably, which costs exactly the same and hurts more.

What survives 🛡️

The ground worth standing on when the questions start.

• The underlying gap is real and worth solving — victims genuinely do lose contact with support services during long trials, and a reliable follow-up and case-continuity system addresses that without needing to infer anyone's mental state
• Asking people directly what they need, in their own language, on a channel that works on a feature phone, is both more accurate than inference and far more respectful, and it is entirely buildable
• A team that reframes this from prediction to accessible follow-up and explains the reasoning is making an argument a panel from this department is well placed to appreciate