DEV Community

Cover image for Nari : Woman Health Meet AI
Saurabh Singh
Saurabh Singh

Posted on

Nari : Woman Health Meet AI

This is a submission for the Google AI Studio Multimodal Challenge

I built a Women’s Health AI Web App that focuses on improving early breast cancer detection, overall women’s health awareness, and doctor efficiency.

The app solves two main problems:

For Doctors – It reduces workload by automatically triaging mammograms and other scans, generating annotated heatmaps, risk scores, BI-RADS classifications, and longitudinal comparisons. Doctors receive AI-prepared reports that they can manually review, saving significant time.

For Patients – It empowers women with personalized education and health guidance. The app generates simplified visual reports, self-exam teaching modules, cycle tracking, pregnancy visuals, nutrition annotation, and skin lesion checks.

This creates a dual experience:

Doctors get a technical clinical assistant that improves efficiency.

Patients get a personalized health companion that simplifies complex medical data.
Uploading a mammogram and receiving an annotated heatmap with BI-RADS score.

Patient-friendly report with plain-language explanation.

Doctor dashboard with Smart Triage Queue and longitudinal tumor tracking.

Teaching module showing self-breast exam steps generated as annotated visuals.

(If Gemini 2.5 Flash Image trial access is no longer active, include a demo video showcasing the AI-generated annotations, risk scoring, and education modules.)

🛠️ How I Used Google AI Studio

I leveraged Google AI Studio as the foundation for my app by integrating Gemini’s multimodal capabilities:

Text + Image Understanding → Mammogram, ultrasound, and X-ray scans are uploaded, and Gemini interprets the medical content, returning clinical risk scores and explanations.

Image Annotation → Gemini automatically highlights suspicious areas in scans (tumors, lesions, cysts, bone density loss).

Image Generation → Used for patient education (e.g., breast self-exam visuals, pregnancy growth diagrams, healthy vs unhealthy mole comparisons).

Multimodal Context Handling → The model takes into account both the image and patient history (age, symptoms, family risk factors) to generate personalized outputs.

This allowed me to build a system where Gemini is not just analyzing but also teaching, guiding, and supporting clinical workflows.

🌐 Multimodal Features

Here are the key multimodal features and why they matter:

Scan Upload + Annotation (Image → Text + Image)

Doctors and patients upload medical scans.

Gemini analyzes and returns annotated heatmaps, BI-RADS categories, and explanations.

Enhances accuracy, speed, and clarity for doctors.

Report Generation (Text → Dual Reports)

Clinical report (technical, BI-RADS, risk factors).

Patient report (plain language, visuals).

Enhances communication between doctors and patients.

Teaching Modules (Image Generation)

AI creates step-by-step annotated visuals for breast self-exam, menstrual cycle, pregnancy stages, bone health, and skin checks.

Enhances health literacy and self-awareness among women.

Doctor Efficiency Tools (Multimodal Decision Support)

Smart triage: AI sorts critical scans first.

Longitudinal tracking: Gemini compares old and new scans.

Enhances doctor productivity and clinical outcomes.

Telemedicine Integration

Remote patients can upload scans.

AI pre-screens before sending to doctor.

Enhances accessibility for rural/low-resource areas.

In summary:
This app leverages Gemini multimodal capabilities to build a full-stack women’s health assistant that helps doctors detect breast cancer faster and educates women on their health journey through visual, annotated, and personalized experiences.

app link : https://ai.studio/apps/drive/1jczl5FeK5f81lG3ekBkOoEKiEIHe-Zyr

Top comments (0)