Objective: This study developed an adaptive short form of the Tinnitus Handicap Inventory (THI) using a Genetic Algorithm (GA) to facilitate online tinnitus treatment triage. The full 25-item THI can increase patient burden, highlighting the need for a shorter, effective alternative.
Design: A genetic algorithm (GA) selected THI items in a three-stage scheme (THI-12/8/5). The GA was trained on Dutch THI datasets and tuned to minimise the difference between projected and full THI-25 scores below the mild threshold.
Study sample: The GA was trained on 1,121 questionnaires and validated on a set of 1,181 questionnaires. Cross-validation was performed using a Polish THI-dataset. Participants, aged 18+, originated from two Dutch hospitals and consented to the use of their data for research.
Results: The analysis produced the THI-12/8/5 model, with 12 items identifying "slight" tinnitus and an additional 8 items for "mild" cases. Patients with higher predicted burdens completed the full 25-item THI. The abbreviated questionnaire demonstrated clinically relevant accuracy while reducing the response burden for less severe cases.
Conclusions: The adaptive THI-12/8/5 is a concise and effective self-report tool for online tinnitus treatment triage. It improves efficiency, reduces patient burden, and contributes to the advancement of abbreviated questionnaires using GA models.
扫码关注我们
求助内容:
应助结果提醒方式:
