Carely

Framing the Problem

Many elderly people struggle with managing healthcare communication. They may forget to take important medication, miss doctor appointments, or have difficulty reaching caregivers quickly when they need help.

At the same time, elderly people are frequently targeted by scam calls that attempt to steal personal or financial information. These scams can be difficult to detect and can cause serious financial and emotional harm.

These issues mainly affect elderly individuals and their caregivers, who need safer and clearer communication tools to manage healthcare tasks and protect themselves from fraud.

Carely aims to address this problem by improving how elderly users communicate with healthcare reminders, caregivers, and incoming callers using AI-powered communication tools.

Idea Explanation

Carely is an AI-powered healthcare companion designed to improve communication and safety for elderly users.

Our system combines AI with communication tools from Twilio to create a safer and more supportive communication experience.

Carely helps solve the problem by providing three key features:

• AI scam call screening - incoming calls are recorded and analysed using AI to determine whether they are legitimate or potentially fraudulent, helping protect elderly users from scam calls.

• Medication reminders - users can create reminders through a simple interface and receive SMS reminders so they do not forget important health tasks.

• Emergency caregiver contact - users can quickly call or send an alert to a caregiver if they need help.

By combining AI analysis and communication tools, Carely helps elderly users stay safe, remember important health tasks, and remain connected with caregivers.

Implementation

Carely is built using a full-stack web architecture with a React frontend and a Node.js backend.

Frontend:

• Built using React

• Provides an accessible interface where users can create reminders, view reminders, and contact caregivers.

Backend:

• Built with Node.js and Express

• Handles reminder scheduling, call processing, and communication with external APIs.

Communication Flow:

1. The user creates a reminder through the React interface.

2. The frontend sends a request to the backend API.

3. The backend schedules the reminder and triggers communication services when the time arrives.

Twilio Integration:

• The Twilio Voice API handles incoming calls and records the caller's message.

• The recorded audio is analysed using AI models (Gemini or OpenAI) to classify whether the call is legitimate or a scam.

• The Twilio Messaging API is used to send SMS reminders for medication and other health-related tasks.

Challenges

One of the biggest challenges was integrating Twilio messaging with our backend in a limited hackathon timeframe.

While implementing SMS reminders, we encountered Twilio trial account restrictions that prevented messages from being sent to unverified international phone numbers. Although our system successfully scheduled and triggered reminders, the Twilio account configuration blocked message delivery.

To diagnose this issue, we verified that the entire technical pipeline was working correctly:

Frontend → Backend → Scheduler → Twilio API

Another challenge was integrating AI-based call screening. This required capturing recorded voice messages from Twilio and sending them to external AI models for classification.

By debugging the API integration and testing each component separately, we were able to successfully implement working call screening and reminder scheduling.

Accomplishments

During the hackathon we successfully:

• Built a full-stack healthcare communication platform

• Implemented AI-based scam call screening

• Integrated Twilio Voice API for call handling

• Implemented reminder scheduling and SMS messaging via Twilio

• Created a clean and accessible user interface for elderly users

Through this project we learned how to integrate communication APIs, connect frontend and backend systems, and use AI models to analyse real-world data such as voice recordings.

Most importantly, we demonstrated how AI and communication technology can work together to improve safety and independence for vulnerable users.

Next Steps

Future improvements for Carely include:

• Expanding SMS reminders to support recurring daily and weekly reminders

• Improving AI scam detection using larger datasets and more advanced models

• Adding voice-based reminder creation to improve accessibility

• Adding a check-in feature so caregivers can confirm that users are safe and have completed important tasks

• Introducing additional accessibility settings (such as larger text, simplified interfaces, and voice assistance) to better support elderly users

• Integrating with healthcare systems such as the NHS, allowing reminders to automatically sync with patient appointments and prescriptions while ensuring strict compliance with data protection and GDPR regulations

• Expanding the platform so that Carely can support users of different age groups, not just elderly individuals, making it a broader healthcare communication assistant

• Enhance the SOS feature so that it can automatically identify the nearest family member or caregiver based on location, allowing the user to quickly contact the closest person in case of an emergency

Our long-term goal is to create a platform that empowers elderly users to manage healthcare communication safely and independently.