Page 125 - AI for Good Innovate for Impact
P. 125
AI for Good Innovate for Impact
f� Setup Reference tools, Notebooks and Simulation Environment
A standardized model simulation and testing environment will be established, along with the
creation of an ophthalmic AI diagnostic simulation platform to test and optimize real-world
medical scenarios. Provide detailed technical documentation and operation instructions to 4.1-Healthcare
lower the technical entry barrier.
3 Use Case Requirements
REQ-01 (Functional): The system must accurately analyze diverse ophthalmic imaging data (e.g.,
fundus photographs, OCT scans) to detect key pathological signs associated with common
preventable or treatable blinding eye diseases such as Diabetic Retinopathy (DR), Glaucoma,
Age-related Macular Degeneration (AMD), and Pathological Myopia.
REQ-02 (Functional): The system must provide clinicians with clear, evidence-based outputs,
including diagnostic likelihoods, disease severity grading (where applicable), identification
of critical findings requiring urgent referral, and potentially risk stratification for disease
progression.
REQ-03 (Functional): The system must support multi-modal data input, capable of processing
not only images but also relevant clinical text information (e.g., patient history snippets,
symptoms described) to provide a more contextualized assessment.
REQ-04 (Functional): The system must offer a natural language interface allowing clinicians to
interactively query for differential diagnoses, ask specific questions about findings, or retrieve
relevant information pertaining to the case, mimicking consultation with an expert.
REQ-05 (Operational): The system must ensure robust data privacy and security, complying
with relevant healthcare data protection regulations through measures like data anonymization,
encryption, and secure access controls.
REQ-06 (Technical): The underlying AI models must be rigorously validated on diverse,
representative datasets for accuracy, fairness, and robustness across different populations
and equipment types.
89
