Scholarlink

Background <p>Cancer-related pain is a multidimensional phenomenon, and key determinants of pain intensity remain unclear. The aim of this study was to identify the clinical determinants of cancer-related pain intensity using a convergent, multimethod analytical framework.</p> Methods <p>Prospective observational study of patients with cancer-related pain. Pain was assessed using a 0–10 numeric rating scale (NRS), the Douleur Neuropathique 4, and the Brief Pain Inventory (BPI). Clinical variables included Eastern Cooperative Oncology Group Performance Status (ECOG-PS), metastatic status, comorbidities, and treatments. Multivariable regression, unsupervised clustering, and supervised machine-learning (ML) models were applied.</p> Results <p>Eighty-six patients were included (median age 64&#xa0;years). In bivariate analyses, higher pain intensity was associated with opioid use and poorer functional status, while neuropathic pain features were prevalent but not independently associated with pain intensity. Spearman correlation analysis demonstrated significant associations between ECOG-PS and pain intensity, BPI severity, and BPI interference. In multivariable regression, ECOG-PS emerged as a key independent predictor of pain intensity, whereas metastatic status and cancer type were not independently associated. Unsupervised clustering identified two clinically meaningful phenotypes primarily distinguished by pain intensity and functional interference, rather than neuropathic mechanisms. In supervised ML models predicting high pain burden (NRS ≥ 7), ECOG-PS consistently ranked as the most influential feature, although overall predictive performance was modest.</p> Conclusions <p>Functional impairment represents a central determinant of cancer-related pain intensity, surpassing pain phenotype and disease-related variables across complementary analytical approaches. These findings support function-centered pain assessment frameworks and underscore the need to systematically integrate functional evaluation into cancer pain management strategies.</p> Trial registration <p>The study was prospectively registered on ClinicalTrials.gov (identifier: NCT07038434).</p>

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

A prospective multimethod investigation of cancer-related pain integrating clinical data and machine learning: results from the RUGGI Study

  • 期刊论文

  • 出版模式：

    开放获取

  • 发表日期：

    2026年4月30日

查看全文
  • Marco Cascella,
  • Cosimo Guerra,
  • Rosario De Feo,
  • Flavio Di Lisio,
  • Pierpaolo Giordano,
  • Walter Esposito,
  • Gennaro Cisale,
  • Valentina Cerrone,
  • Dalila Esposito,
  • Maria Pia Bruno,
  • Rossella Tarallo,
  • Martina Lombardi,
  • Jacopo Troisi,
  • Marzio Galdi,
  • Stefano Martina,
  • Adele Zarrella,
  • Enza della Rocca,
  • Amelia Filippelli,
  • Valeria Conti,
  • Mario Montedoro,
  • Francesco Sabbatino,
  • Giuseppe Polese,
  • Ornella Piazza

摘要

Background

Cancer-related pain is a multidimensional phenomenon, and key determinants of pain intensity remain unclear. The aim of this study was to identify the clinical determinants of cancer-related pain intensity using a convergent, multimethod analytical framework.

Methods

Prospective observational study of patients with cancer-related pain. Pain was assessed using a 0–10 numeric rating scale (NRS), the Douleur Neuropathique 4, and the Brief Pain Inventory (BPI). Clinical variables included Eastern Cooperative Oncology Group Performance Status (ECOG-PS), metastatic status, comorbidities, and treatments. Multivariable regression, unsupervised clustering, and supervised machine-learning (ML) models were applied.

Results

Eighty-six patients were included (median age 64 years). In bivariate analyses, higher pain intensity was associated with opioid use and poorer functional status, while neuropathic pain features were prevalent but not independently associated with pain intensity. Spearman correlation analysis demonstrated significant associations between ECOG-PS and pain intensity, BPI severity, and BPI interference. In multivariable regression, ECOG-PS emerged as a key independent predictor of pain intensity, whereas metastatic status and cancer type were not independently associated. Unsupervised clustering identified two clinically meaningful phenotypes primarily distinguished by pain intensity and functional interference, rather than neuropathic mechanisms. In supervised ML models predicting high pain burden (NRS ≥ 7), ECOG-PS consistently ranked as the most influential feature, although overall predictive performance was modest.

Conclusions

Functional impairment represents a central determinant of cancer-related pain intensity, surpassing pain phenotype and disease-related variables across complementary analytical approaches. These findings support function-centered pain assessment frameworks and underscore the need to systematically integrate functional evaluation into cancer pain management strategies.

Trial registration

The study was prospectively registered on ClinicalTrials.gov (identifier: NCT07038434).