Service Plan

Questions	Answers
평가 기준에서 중점적으로 고려하시는 사항이 무엇인지요?	문의해주신 평가 기준은 아래와 같습니다. ● General impact on Healthcare ○ Clearly defined unmet medical needs and the goals and functions of the proposedsolution to overcome those.○ Potential Impact/scalability of the proposed solution for improving Korean healthcare.○ Potential impact/scalability/innovativeness of the proposed solution for improvingU.S./Global healthcare. ● Product Readiness ○ Previous demonstration of the solution's capability using Korean health data (i.e.proof of concept).○ Technology assessment (Is this a novel approach with competitive advantages?). ● Validation Feasibility of proposed work by the UCLA Center for SMART Health (referred to as UCLA) ○ Compatibility with UCLA validation platform (i.e., containerization of the digitalsolution and general availability and standardization of input data).○ Concrete definition of the input data for the solutions and the output metric (e.g.,AUROC, F1 score, etc.) 지원서에 위의 평가 기준에 맞추어 세부항목을 마련해 놓았으니 참고하여 작성 부탁드립니다.
UCLA 의료데이터를 이용한 검증은 구체적으로 어떻게 진행되는지요?	우승팀이 출품한 솔루션은 docker 컨테이너화 하여 black-box 형태로 UCLA 스마트 헬스 센터 측으로 전달합니다. 또한, 우승팀이 솔루션의 입력 데이터에 대한 명확한 정의와 구성을 안내하면 UCLA측에서는 관련 의료데이터를 확보하여 컨테이너화 된 솔루션에 적용하여 예측값을 산출합니다. 이 결과는 예측을 평가하는 다양한 지표로 우승팀에 제공됩니다. 다음은 UCLA측에서 제공한 검증에 대한 가이드라인입니다. 검증과정 동안 우승팀과 UCLA측 간의 소통을 통해 추가적인 조율이 가능합니다. The winning team shall provide a solution that can be deployed following UCLA’s specs for the validation process which are: The solution will be hosted in Amazon’s AWS cloud UCLA infrastructure and more particularly in an EC2 Ubuntu machine with the following specs. Ubuntu 20.04 amd64 instance type as t2.xlarge (4vCPUs and 16GB memory) EBS Volume size to 250GB The solution will be installed in a Docker container and the UCLA CTSI team operates it without interacting with the inner workings and components of the model as in following a black box approach. The winning team shall be responsible for providing instructions to deploy their containerized solution without directly participating in it (they will not have direct access to UCLA’s validation environment). Following validation, the winner will receive a validation report that shall include a comprehensive statistical analysis done by UCLA personnel. De-identified UCLA data will be used for this purpose but no patient data will be shared with the winner, only the results of the analysis.