AI disease Diagnosis & treatment suggestion App
Budget: ₹37,500 – ₹75,000 INR
I’m building an app that diagnoses diseases and predicts/suggest treatments and we need an AI specialist who can turn raw clinical data into fast, reliable diagnoses. if you already have access to clinical data, it will be an added advantage. The first release must confidently handle Diabetes and Cancer.
The workflow I have in mind starts with a symptom-assessment module that collects patient-reported information and any available vitals. From there, a risk-scoring layer weighs lifestyle and genetic inputs to triage urgency and feed a recommendation engine. Finally, I want the system to interpret PET-scan imagery so we can cross-validate the text-based findings and surface personalised health recommendations the moment results are ready.
Key deliverables
• Train and deploy models (or fine-tune existing ones) that flag Diabetes, Heart Disease, and Cancer with high precision and recall.
• Build a symptom-assessment interface that maps free-text or structured inputs to ICD-10 terminology.
• Integrate PET-scan analysis—OpenCV, TensorFlow or PyTorch are all acceptable—as an optional but seamless step in the diagnostic flow.
• Generate personalised lifestyle and treatment advice following evidence-based guidelines, clearly separating AI suggestions from clinician notes.
• Package everything behind a REST or GraphQL API so a mobile front-end team can plug in without extra tooling.
Acceptance criteria
• AUROC ≥ 0.90 for each disease on an unseen test set.
• End-to-end inference time under 3 s on a mid-range cloud instance.
• Codebase containerised (Docker) with reproducible training scripts and a short README.
If this tech stack and scope fit your expertise, let’s talk timelines and milestones.
The workflow I have in mind starts with a symptom-assessment module that collects patient-reported information and any available vitals. From there, a risk-scoring layer weighs lifestyle and genetic inputs to triage urgency and feed a recommendation engine. Finally, I want the system to interpret PET-scan imagery so we can cross-validate the text-based findings and surface personalised health recommendations the moment results are ready.
Key deliverables
• Train and deploy models (or fine-tune existing ones) that flag Diabetes, Heart Disease, and Cancer with high precision and recall.
• Build a symptom-assessment interface that maps free-text or structured inputs to ICD-10 terminology.
• Integrate PET-scan analysis—OpenCV, TensorFlow or PyTorch are all acceptable—as an optional but seamless step in the diagnostic flow.
• Generate personalised lifestyle and treatment advice following evidence-based guidelines, clearly separating AI suggestions from clinician notes.
• Package everything behind a REST or GraphQL API so a mobile front-end team can plug in without extra tooling.
Acceptance criteria
• AUROC ≥ 0.90 for each disease on an unseen test set.
• End-to-end inference time under 3 s on a mid-range cloud instance.
• Codebase containerised (Docker) with reproducible training scripts and a short README.
If this tech stack and scope fit your expertise, let’s talk timelines and milestones.