MEDIGATE-RX

• 

  here, the user asks about a general health question, the system provide general knowledge without restraint

• 

  user ask for treatment for tonsillitis, the system detects this as a self- treatment intent question, and directs user to a doctor

• 

  user asks about a question on online trading, which will be classified as non risky query, the system provide answers without restraint

• 

  user uploads an image of a condition he is having, the system detects this a self diagnosis intent question and directs user to a doctor

Inspiration

We are living in an era of AI, where each developer wants to make a smarter AI, one that provides answers to every question asked, without considering safety issues that might arise, especially in the healthcare system. So, current AI platforms, if a person asks a question about their medical condition, be it self-diagnosis or possible treatments for their health condition, the answer is always provided, then the system will give a disclaimer and suggest consulting a healthcare professional. By this point, however, the user's question has already been addressed—potentially encouraging self-diagnosis or self-treatment. So now, considering also the rise of the pharma black market, where medications are not being sold in registered or authorized premises, people are getting them without prescriptions, some of them are not of standard quality, counterfeit drugs, antibiotics and drugs which are contain toxic materials used to mimic the appearance or physical effects of the real drug, such as mercury, arsenic. This combination alone is a major driver of antimicrobial resistance and adverse drug reactions, affecting the well-being of those involved and placing a significant burden on global health systems, such as the World Health Organization (WHO). This insight directly influenced MEDIGATE-RX's mission to prevent self-diagnosis and unsafe self-medication, ensuring public well-being.

What it does

MEDIGATE-RX is a clinical safety system that acts as a pre-processing layer for AI applications. It analyzes user queries before any response is generated and classifies them based on their risk levels.

For non-clinical queries, the system responds normally. For general health questions, it provides safe educational information. However, when a query indicates self-diagnosis or self-treatment, the system blocks unsafe outputs and redirects the user to a healthcare professional.

By preventing harmful clinical guidance at the source, MEDIGATE-RX improves the safety of AI-assisted healthcare and can be integrated into existing chatbots and telehealth platforms.

How we built it

I built MEDIGATE-RX using Google AI Studio by designing a prompt-based system that simulates a clinical safety layer. The system consists of two main components: a risk classifier and a controlled response generator.

The classifier analyzes user queries and assigns a risk level based on whether the query is non-clinical, general health, or related to self-diagnosis or self-treatment. Based on this classification, the response generator applies strict safety rules to either allow, limit, or block the output.

The entire system was implemented using structured prompt engineering, without traditional coding, and demonstrated through real-time interaction in Google AI Studio.

Challenges we ran into

Since I was using Google Studio AI free tier, I had to make sure I did not exhaust my quota

Accomplishments that we're proud of

I am proud of designing a functional clinical safety system using only prompt engineering, without traditional coding. I successfully created a structured framework that classifies user queries by risk and controls AI responses accordingly.

I am also proud of addressing a real-world healthcare problem by preventing unsafe self-diagnosis and self-medication behaviors, which are linked to antimicrobial resistance and adverse drug reactions.

What we learned

What's next for MEDIGATE-RX

The next step is to develop MEDIGATE-RX into a fully integrated application that can connect with real-world chatbot and telehealth systems.

Future improvements include enhancing the accuracy of the risk classification model, incorporating real-time medical guidelines, and expanding the system to handle more complex clinical scenarios.

Built With

• chatgpt
• gemini
• google-ai-studio