Inspiration
In a hospital, every minute matters. Yet a significant part of a doctor's time can be spent collecting patient history, reviewing fragmented records, repeating questions, and manually documenting consultations.
We were inspired by the problem of overcrowded healthcare environments, particularly where large patient volumes, paper-based records, language barriers, and detailed AYUSH history requirements make clinical intake slower and more difficult. Traditional systems can result in incomplete histories and information being lost between different records and providers.
This led us to a simple question: What if the patient's history could be collected, organized, and prepared before the doctor has to spend valuable consultation time doing it?
That idea became MediKiosk, our approach to making patient case-taking faster, more structured, and more accessible.
What it does
MediKiosk is an AI-powered patient case-taking platform that assists patients and healthcare workers before and during a consultation.
It provides:
Multilingual AI-guided interviews through voice and touch. Adaptive questioning based on the patient's responses. Red-flag screening to identify potentially urgent cases and alert healthcare staff. OCR-based document processing for prescriptions, reports, and previous medical records. Voice-to-EMR documentation during doctor-patient consultations. AI-generated clinical summaries that doctors can review. AYUSH-specific case-taking, including workflows for detailed traditional medicine history. Offline-first operation for environments with limited connectivity. ASHA/ANM-assisted mode for patients who may need help using the system. A path toward ABHA/ABDM and FHIR-based integration with digital health systems.
The goal is not to replace the doctor. It is to remove repetitive documentation and information-gathering work so that the doctor can spend more of the consultation actually treating the patient.
How we built it
We designed MediKiosk as a modular architecture where different AI capabilities work together rather than relying on a single model.
The workflow starts with registration, consent, language selection, and red-flag screening. The patient then goes through an AI-guided interview covering symptoms, medical history, medications, family history, AYUSH information, and previous records.
The collected information passes through components for speech recognition, translation, OCR, clinical analysis, summarization, and structured record generation. The final information is presented through a doctor dashboard, where the doctor can review and edit the AI-generated output before it becomes part of the medical record.
We also adopted an offline-first approach. Core functionality is designed to continue working without continuous internet connectivity, with data synchronization when connectivity becomes available. This makes the system more suitable for low-connectivity healthcare environments.
Challenges we ran into
The biggest challenge was realizing that healthcare AI cannot afford to behave like ordinary AI.
A hallucinated sentence in a chatbot is frustrating. A hallucinated symptom in a medical record can be dangerous.
We therefore designed the system so that it does not fill missing information with assumptions. Unconfirmed information remains blank, while AI-generated information is treated as a suggestion that requires human verification.
We also faced several practical challenges:
Noisy OPD environments can reduce speech-recognition accuracy. Mixed-language speech can make transcription and translation difficult. Medical translation requires preserving the meaning of symptoms and clinical terminology. Blurry or handwritten prescriptions can produce OCR errors. Poor connectivity can interrupt cloud-based services and synchronization. Low digital literacy means that a conventional app interface cannot be assumed to be accessible to every patient.
These challenges shaped the product itself. Voice has a touch-based fallback, OCR retains the original document, unclear information can be flagged, and doctors remain in control of the final record.
Accomplishments that we're proud of
We are proud that MediKiosk evolved beyond the idea of simply putting an AI chatbot in a hospital.
We created a complete patient-intake workflow that connects patient interaction, clinical history, document processing, AI analysis, doctor review, and digital records.
Some of the aspects we are particularly proud of are:
Combining voice + touch interaction to make the system accessible. Designing for 22+ Indian languages through multilingual AI capabilities. Building an offline-first architecture rather than assuming constant internet access. Incorporating red-flag screening into the intake process. Creating a dedicated AYUSH workflow instead of forcing traditional medical history into a generic template. Keeping the doctor in control of AI-generated information. Designing toward FHIR and ABDM compatibility for future interoperability. Creating an ASHA/ANM-assisted mode to extend the system beyond digitally confident users.
What makes us most proud is the philosophy behind the system: AI should carry the administrative burden, not the clinical responsibility.
What we learned
The project taught us that building healthcare technology is less about asking, “What can AI do?” and more about asking, “What should AI be trusted to do?”
We learned the importance of human-in-the-loop systems, especially when dealing with medical information. We also learned that accessibility cannot be added at the end of development. Language, literacy, connectivity, and the realities of a crowded hospital environment have to influence the architecture from the beginning.
Most importantly, we learned that the difficult part is not making information digital.
The difficult part is making it reliable enough to matter.
What's next for Patient Case-Taking Software
MediKiosk is designed to grow in stages.
Our immediate goal is to pilot the system in PHCs and AYUSH centers, where we can validate the workflow in real healthcare environments. From there, we plan to add ASHA/ANM mobile support, connect with existing EMR/HIS systems through FHIR, and enable ABHA/ABDM integration when connectivity is available.
The longer-term vision is to expand from individual healthcare centers into larger hospital and government healthcare systems.
We want MediKiosk to become a quiet layer between the patient and the paperwork, gathering the story that might otherwise be lost in queues, handwritten pages, repeated questions, and fragmented records.
Because ultimately, a medical consultation begins with a story.
Our job is to make sure that story is heard, understood, and carried forward.
Log in or sign up for Devpost to join the conversation.