Polly

Inspiration

The inspiration for our project came from a mutual interest in women's health. When we met, it was one of the first topics we discussed and we immediately decided to choose it as our topic. There is a noticeable lack of support and research in relation to women's health, and we knew that, as STEM students, we were in position to apply our knowledge to help change this.

What it does

We propose a supportive device and app system for women with insulin-resistant PCOS that focuses on: a) AI/ML-supported insight -A privacy-preserving personalized prediction and recommendation system that combines local user-specific model adaptation with Retrieval-Augmented Generation (RAG) for evidence-grounded outputs. This ensures information is tailored to the individual user and backed by trustworthy and verified sources. -An AI chatbot that explains patterns in simple language, connects glucose trends with lifestyle, symptoms, and medication changes and supports reflection rather than judgement.

b) Lifestyle and medication support -Personalised diet and exercise guidance based on goals (weight management, symptom control, fertility, energy), restrictions and preferences and clinician input where appropriate. -Medication timelines to help users and doctors see what changed, when it changed and whether benefits outweigh side effects. This system does not replace clinicians, it supports better conversations and decision-making.

How we built it

We used figma to create the prototype of the app. We used Copilot to generate the images.

What to expect

Our app will provide users with accurate, personalised information describing typical symptoms during each stage of the menstrual cycle for those with PCOS. The app will learn from past symptoms inputted by the user including flare-ups and history of triggers including foods that cause insulin spikes and increased inflammation. The app’s insights and suggestions would detail customized advice on recommended actions, for example increasing sleep and prioritising a low-inflammation diet.

Chatbot

Within the app, our AI chatbot Polly aims to provide clear, direct advice tailored to the individual needs of the user. Using information about each user’s medical conditions and history, Polly will generate unique guidance and advice to on diet, lifestyle and XX to allow users to take control over their own symptom management. Over time, Polly will further learn to adapt to each individual user, basing new advice off feedback provided by the users. Polly, along with several other app features, will implement Retrieval-Augmented Generation framework, a technique which allows for the retrieval of data from specific data sources relevant to the topics at hand.

Data

Our app connects to other apps and smart devices the user might have including smart watches (Apple Watch, Fitbit, Garmin, Health App) and insulin monitors via Bluetooth. This will give the app access to someone’s insulin levels, heart rate, sleep, stress levels etc. This gives the app a clearer picture of the user and their needs.

Medication Page

The user can input what medication they are currently on. This page is linked to the daily log. If there are any chances in medication the person can update it and the app will track if this affects how they are feeling.

History and Daily Log

The person can input any other conditions that they might have, for example Ehlers- Danlos Syndrome or fibromyalgia. The app can then inform the user of any effects or other symptoms they should be aware of. There will be an overview of how the user has been feeling, and machine learning will be used to recognise any patterns and analyse previous months and cycles, for example if they feel a certain way in the days leading up to their period. The app can then generate personalised insights and recommendations for the user. The daily log will include: how the person is feeling, pain levels (rated on a scale, type of pain, location), any other symptoms they are currently experiencing, exercise and activities. This will be linked to the medication page and history.

Challenges we ran into

When coming up for an idea for our project, we needed to pick a specific topic to focus on. Some of our early ideas were too broad, and we had to spend some time narrowing down our scope to be a more manageable project.

Accomplishments that we're proud of

Our app could make a real difference in people's lives. We started with an abstract idea and we designed a prototype that might solve a real-world problem. We built a user-centered app focused on accessibility and usability.

What we learned

During our project, we learned how to apply our knowledge from several different fields of knowledge, and work together effectively to create a useful, practical app. We learned how to choose a relevant topic that would make a difference in the world today, and how to focus on the issues that would make the most impact.

What's next for Polly

As Polly progresses, we hope to create similar apps focussing on a wider range of health conditions affecting women such as endometriosis. We also hope to expand the app, improving recommendations and making the experience in the app more personalised for every user.

Built With

• canva
• capcut
• copilot
• figma