DengueOps AI

Responsible AI

Ethics & Responsible Use

“Privacy-safe, human-in-the-loop decision support for dengue preparedness.”

DengueOps AI is designed as a public health preparedness prototype, not a diagnostic or autonomous decision-making tool. It uses aggregated and synthetic demonstration data to show how outbreak forecasts can be translated into readiness alerts and action priorities without processing patient-level information.

No patient-level dataSynthetic operational dataHuman-in-the-loopAdvisory outputsNo diagnosisTransparent assumptions

Design intent: This prototype demonstrates a privacy-safe, transparent, and human-in-the-loop approach to AI-enabled public health preparedness. It prioritises responsible simulation over unsupported claims of live deployment.


Design Principles

Ethical Design Principles

Six core ethical commitments are embedded in the prototype design.

No Patient-Level Data

The prototype does not collect, store, or process identifiable patient-level records. Forecasting and operational outputs are based on aggregated/demo data structures.

Privacy-Safe Demonstration

Facility readiness, inventory, and dengue-specific bed values are synthetic demonstration values. This prevents unauthorised use of sensitive hospital operations data.

Human-in-the-Loop Decision-Making

Outputs are advisory. Final decisions remain with public health officials, hospital administrators, and qualified authorities.

No Clinical Diagnosis

The system does not diagnose dengue, recommend individual treatment, or replace clinicians.

Transparent Uncertainty

Forecast uncertainty is shown using best, expected, and worst-case scenarios so decision-makers can plan cautiously.

Bias and Vulnerability Safeguards

The vulnerability-gated priority score prevents static vulnerability factors from permanently dominating response priorities when forecasted dengue risk is low.


Data Ethics

Data Ethics and Data Boundaries

Clear boundaries define what is real, what is anchored from public sources, and what is entirely synthetic demonstration data.

Data ElementStatusNote
Dengue case dataSynthetic / DemoDemo aggregate values for pipeline demonstration.
Climate dataSynthetic / DemoSynthetic/demo or public-style aggregate values.
Public hospital namesPublic AnchorUsed as geographic facility anchors where available.
General bed capacityPublic AnchorPublic reference anchor where available; synthetic otherwise.
Dengue-specific bedsSynthetic / DemoSynthetic demonstration values only.
Current dengue occupancySynthetic / DemoSynthetic demonstration values only.
NS1/RDT stockSynthetic / DemoSynthetic demonstration values only.
IV fluid stockSynthetic / DemoSynthetic demonstration values only.
Baseline daily consumptionSynthetic / DemoSynthetic demonstration values only.
Patient-level recordsNot UsedNot collected, processed, or stored.

Exact statement: Facility names and general bed-capacity anchors are based on public/government references where available. Dengue-specific bed allocation, current occupancy, NS1/RDT stock, IV fluid stock, and consumption values are synthetic demonstration values.


Safety Boundaries

What the System Does Not Do

Clear negative constraints are part of the responsible design.

It does not — Diagnose dengue.

It does not — Recommend individual treatment.

It does not — Replace clinical judgment.

It does not — Claim access to live hospital stock or bed data.

It does not — Automatically trigger public health action.

It does not — Provide official public warnings.

It does not — Process patient-identifiable data.

Output Translation

Responsible Output Design

Technical model outputs are translated into operational preparedness metrics before reaching decision-makers.

Technical layer

  • · MAE / RMSE
  • · Forecast uncertainty
  • · Model comparison
  • · Feature lags

Operational layer

  • · Risk level
  • · SDH
  • · Bed gap
  • · Priority score
  • · Recommendations

Decision layer

  • · Human review
  • · Facility planning
  • · Vector-control prioritisation
  • · Contingency planning

RMSE and model metrics are included for technical validation and evaluator transparency; operational users receive translated preparedness outputs rather than raw model diagnostics.


User Roles

User Roles and Ethical Responsibilities

Each user role interacts with the prototype in a defined, bounded capacity.

MIS / Data Officer

  • · Maintains the analytics data pipeline
  • · Validates input files before scheduled runs
  • · Runs scheduled pipeline updates

Public Health Analyst

  • · Reviews forecast behaviour and uncertainty ranges
  • · Checks assumptions and data limitations
  • · Interprets outbreak patterns cautiously

Hospital Administrator

  • · Uses readiness alerts for facility planning
  • · Confirms real stock and bed status before taking action
  • · Does not treat synthetic values as definitive

City Corporation / Vector-Control Team

  • · Uses zone priority score as one input among many
  • · Confirms local field conditions before deployment
  • · Does not act on prototype output alone

Technical Evaluator

  • · Reviews model validation, limitations, and assumptions
  • · Assesses pipeline design and decision-support logic
  • · Does not evaluate as a live clinical system

Future Deployment

Future Deployment Ethics

Real operational deployment of this system would require formal governance and institutional safeguards beyond the prototype scope.

1

Institutional approval and governance

2

Official data-sharing agreements

3

Privacy review and data protection assessment

4

Facility-level validation of readiness logic

5

Access control and role-based permissions

6

Audit logs for all data and forecast actions

7

Regular model monitoring and recalibration

8

Public health governance and accountability chain

9

Clear legal accountability framework

DengueOps AI demonstrates a privacy-safe, transparent, and human-in-the-loop approach to AI-enabled public health preparedness. The prototype prioritises responsible simulation over unsupported claims of live deployment.

Ethics Summary — DengueOps AI · IEEE ICADHI 2025

DengueOps AI — Simulation-Based Dengue Surge Preparedness Decision Support for Dhaka South.  Prototype only. Not for clinical or official public health use.