MedElevator
Войти в исследования International journal of medical informatics · 2025
Machine learning predicts scarring progression in lichen planopilaris: A multidimensional model integrating trichoscopy, vitamin D, and diagnostic timelines.
СтатьяПоддержка исследования: вне правительства СШАНазвание и аннотация: Перевод в очереди
Обработка статьи · этапы, ошибки и управление
Загрузка состояния…
Ошибки и ограничения
История · было → стало
Состояние обновляется каждые 5 секунд, пока блок открыт. Процент показывается только там, где известен объём работы. Пауза останавливает продолжение этапа; текущий запрос к источнику может завершиться.
Seetan Khaled, Khamees Almu'atasim, Yassin Raghad Yousef, Ba-Shammakh Saleh A, Moh'd Amir Majdi
Аннотация
PURPOSE: Lichen planopilaris (LPP) is a devastating inflammatory hair disorder that leads to permanent scarring alopecia, causing profound psychosocial distress and significantly impairing quality of life. It is caused by lymphocytic inflammation and can progress in different ways. Identifying predictors for disease progression is challenging. This uncertainty affects treatment choices and patient results. To create the first risk assessment model for LPP progression based on machine learning, using various clinical, trichoscopic, and immunological factors.
METHODS: This retrospective cohort study examined 312 biopsy-confirmed LPP patients from 2019 to 2024 at a tertiary center. Disease progression was assessed using the LPPAI and the percentage of scalp involvement. Cox regression and interpretable machine learning (gradient boosting with SHAP) identified predictors of progression.
RESULTS: Rapid progressors (n = 89, 28.5 %) showed higher baseline LPPAI (6.8 ± 1.9 vs 4.2 ± 1.7, p < 0.001), diagnostic delays > 12 months (HR = 3.24, 95 %CI:2.11-4.98), and vitamin D deficiency (HR = 2.56, 95 %CI:1.71-3.83). Trichoscopic severity > 20/30 predicted rapid progression (sensitivity 82.4 %). The gradient boosting machine learning model achieved 92 % accuracy (AUC-ROC = 0.92), 82 % sensitivity, and 85 % specificity at the optimal probability threshold. TreeSHAP analysis identified diagnostic delay, baseline trichoscopic severity score, and vitamin D deficiency as the top-ranked predictors. Decision curve analysis confirmed significant clinical utility across relevant risk thresholds.
CONCLUSIONS: Diagnostic delay, vitamin D deficiency, and trichoscopic severity are key modifiable predictors of disease progression. Our validated model enables personalized risk assessment, early intervention, better long-term results, and lower healthcare costs. Using our risk evaluation method could shift LPP management from a reactive stance to a proactive approach, ultimately helping to preserve hair follicles and enhance the quality of life for patients.
METHODS: This retrospective cohort study examined 312 biopsy-confirmed LPP patients from 2019 to 2024 at a tertiary center. Disease progression was assessed using the LPPAI and the percentage of scalp involvement. Cox regression and interpretable machine learning (gradient boosting with SHAP) identified predictors of progression.
RESULTS: Rapid progressors (n = 89, 28.5 %) showed higher baseline LPPAI (6.8 ± 1.9 vs 4.2 ± 1.7, p < 0.001), diagnostic delays > 12 months (HR = 3.24, 95 %CI:2.11-4.98), and vitamin D deficiency (HR = 2.56, 95 %CI:1.71-3.83). Trichoscopic severity > 20/30 predicted rapid progression (sensitivity 82.4 %). The gradient boosting machine learning model achieved 92 % accuracy (AUC-ROC = 0.92), 82 % sensitivity, and 85 % specificity at the optimal probability threshold. TreeSHAP analysis identified diagnostic delay, baseline trichoscopic severity score, and vitamin D deficiency as the top-ranked predictors. Decision curve analysis confirmed significant clinical utility across relevant risk thresholds.
CONCLUSIONS: Diagnostic delay, vitamin D deficiency, and trichoscopic severity are key modifiable predictors of disease progression. Our validated model enables personalized risk assessment, early intervention, better long-term results, and lower healthcare costs. Using our risk evaluation method could shift LPP management from a reactive stance to a proactive approach, ultimately helping to preserve hair follicles and enhance the quality of life for patients.