Scholarlink

Chronic pain is highly disabling, and it is the main reason to attend rehabilitation. Currently no tool characterises the whole pain experience. The WORDSforPAIN project aims to develop and validate a pain assessment model based on Artificial Intelligence (AI); this is an update on model vadidation. A prospective study, funded by Tuscany Health Ecosystem (PNRR—NewGenerationEU), is conducted on 200 persons with chronic spinal pain. Participants provide brief written stories about their pain experience and undergo a comprehensive biopsychosocial assessment, twice at baseline (T0–T1), and post-rehab (T2); outcomes are also assessed at 6 months follow up by phone. The structure of pain experience and linguistic markers of different pain dimensions are obtained by manual annotation of the texts to define the new pain assessment model and train an AI algorithm. An extensive assessment of construct validity was carried out by computing correlations between presence/absence of pain dimensions and biopsychosocial variables at T0 and T2. To date, 254 texts from 107 participants have been annotated. Construct validity assessment considering 158 diverse experiences (i.e. excluding texts at T2 if participants did not subjectively change compared to T0) reveals: consistent significant weak to moderate associations between pain dimensions and demographic-clinical characteristics (rpb from 0.158 to 0.226), pain features/physical tests (rpb from 0.157 to 0.338), disability/health-related quality of life (rpb from 0.159 to 0.349), and psychological/sleep impairments (rpb from 0.157 to 0.341). In conclusion, the emerging qualitative pain assessment model seems to be sustained by several consistent correlations regarding its construct validity; however, these results must be considered carefully because of study limitations. Completing the enrolment and finalising the study is mandatory to obtain all needed data to reach project objectives.

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Construct Validity of a New Pain Assessment Model: Update on the WORDSforPAIN Study

  • 会议论文

  • 发表日期：

    2026年

查看全文
  • Mario De Marco,
  • Letizia Cipriani,
  • Giulio Cherubini,
  • Marco Baccini,
  • Lorenzo Giotti,
  • Francesco Gariboldi,
  • Alice Livoti,
  • Maria Luigia del Vicario,
  • Stefano Giuseppe Doronzio,
  • Gianmarco Tuccini,
  • Goffredo Guidi,
  • Silvano Zipoli Caiani,
  • Roberta Lanfredini,
  • Francesca Cecchi

摘要

Chronic pain is highly disabling, and it is the main reason to attend rehabilitation. Currently no tool characterises the whole pain experience. The WORDSforPAIN project aims to develop and validate a pain assessment model based on Artificial Intelligence (AI); this is an update on model vadidation. A prospective study, funded by Tuscany Health Ecosystem (PNRR—NewGenerationEU), is conducted on 200 persons with chronic spinal pain. Participants provide brief written stories about their pain experience and undergo a comprehensive biopsychosocial assessment, twice at baseline (T0–T1), and post-rehab (T2); outcomes are also assessed at 6 months follow up by phone. The structure of pain experience and linguistic markers of different pain dimensions are obtained by manual annotation of the texts to define the new pain assessment model and train an AI algorithm. An extensive assessment of construct validity was carried out by computing correlations between presence/absence of pain dimensions and biopsychosocial variables at T0 and T2. To date, 254 texts from 107 participants have been annotated. Construct validity assessment considering 158 diverse experiences (i.e. excluding texts at T2 if participants did not subjectively change compared to T0) reveals: consistent significant weak to moderate associations between pain dimensions and demographic-clinical characteristics (rpb from 0.158 to 0.226), pain features/physical tests (rpb from 0.157 to 0.338), disability/health-related quality of life (rpb from 0.159 to 0.349), and psychological/sleep impairments (rpb from 0.157 to 0.341). In conclusion, the emerging qualitative pain assessment model seems to be sustained by several consistent correlations regarding its construct validity; however, these results must be considered carefully because of study limitations. Completing the enrolment and finalising the study is mandatory to obtain all needed data to reach project objectives.