报告详情
SportAdmit: A Feature-Token Transformer for Hospital Admission Risk after Sports and Recreation Injuries in the NEISS Database
编号：51
访问权限：仅限参会人
更新：2026-11-10 16:31:58 浏览：8次
In-person
报告开始：2026年11月20日 14:45（Asia/Kolkata）
报告时间：15min
所在会场：[S4] Artificial Intelligence and Intelligent Computing [S4-1] Artificial Intelligence and Intelligent Computing
暂无文件
摘要
Sports injuries leading to hospital stays take up a lot of clinical effort even when they are not that frequent overall. The main national data sets still get used mostly for basic counts. We wanted to check if a feature token transformer could sort the cases by risk of admission or transfer or observation. It would use only the coded fields already collected at the emergency department from the 2025 NEISS records.There were seventy four thousand five hundred twenty one sports and recreation cases across fifty five activity types. A bit over five percent of them involved admission transfer or observation. The data got split seventy fifteen fifteen and a thirty one thousand parameter transformer was trained next to a light gbm baseline on the same parts. On the held out cases the transformer showed better performance than the baseline at separating higher risk cases.Top ten percent of scores picked up around sixty percent of the actual admissions. Rates differed quite a bit across sports and also rose with age. It seems some activities carry higher chances than others and older age groups stand out.Because class weights were used during training the output scores work for ranking but are not direct probabilities. That part is easy to miss when looking at the numbers. Some people might expect calibrated outputs right away but the setup was mainly about ordering the cases.
关键词
sports injury, hospital admission, injury surveillance, NEISS, tabular Transformer, gradient boosting
报告人
Dr.Praveen A
PROFESSOR Kerala；Kannur University稿件作者
全部评论
重要日期
-
会议日期
11月20日
2026
至11月22日
2026
-
10月31日 2026
初稿截稿日期
-
11月07日 2026
提前注册日期
-
11月22日 2026
注册截止日期
主办单位
United Societies of Science
承办单位
Pondicherry University
协办单位
IEEE Section
IEEE Madras Section
IEEE Madras Section
联系方式
- ICon INDIA 2026
- ic******@usssociety.org
发表评论