Target for self-developed AI advancement Recruitment of a multi-agency expert evaluation team
Clinical Nutrition AI company Newgem Lab announced that it will introduce the concept and development direction of clinical nutrition vertical AI “NUSEUM AI” currently being developed at HBIC Medi-Tech Demo Zone on the first floor of Korea University Guro Hospital’s Future Hall from the 17th to the 28th
This exhibition is a research exhibition in the development stage to introduce the clinical nutrition data, safety judgment logic, evidence search structure, and AI verification system being built by Newgium Lab and to collect opinions from medical and nutrition experts
Newgem AI is being developed with the aim of providing food, menu, diet, and nutrition guides that are appropriate for individuals based on health status, disease, allergy, medication information and health goals. In the future, Newgium Lab’s goal is to expand into API and SDK-based infrastructure that can be linked to medical, digital health, food and commerce services
NewGeumLab is combining search technology with clinical nutrition knowledge to compensate for inaccurate information and safety issues that may arise in the process of probabilistically generating answers by general Generative AI. It first checks taboo, allergy, and drug and nutritional interactions by disease, searches related literature in consideration of evidence level, up-to-dateness, and question similarity, and then implements a structure in which the reason and source of recommendation are presented together.
In this exhibition, not only the function of AI under development, but also research on the criteria for evaluating personalized clinical nutrition AI will be introduced. The Newgem Lab is embodying the verification system based on the advice and review of researchers from the Korean Nutrition Association and medical institutions
In particular, obesity and GLP-1-based nutrition management are set as priority areas, and a dataset consisting of questions and standard answers reflecting actual clinical and nutritional counseling situations is being established. Questions and answers will be confirmed after expert review and content validity evaluation, and after that, the answers of Newgium AI and the general-purpose giant language model will be blind evaluated under the same conditions
In connection with this exhibition, Newgem Lab will also recruit a multi-organization expert evaluation team. Researchers in related fields such as doctors and clinical nutritionists are recruited. Among the applicants, evaluators will be selected in consideration of their expertise and experience, and the selection status and specific evaluation schedule will be guided individually. The selected evaluator will be involved in evaluating the answers of Newgium AI and comparative AI without the information of the model being disclosed in the future, and reviewing the accuracy, safety, evidence-based and practical utility of personalized clinical nutrition AI.
“This exhibition is a process of disclosing the data, safety logic, and verification criteria needed for clinical nutrition AI to be trusted and seeking the direction of development with experts,” said Kim Ji-young, CEO of NewGium Lab. “We will use this exhibition and recruitment of expert evaluation teams as a starting point for creating objective verification criteria for personalized clinical nutrition AI.”
