Frontiers in immunology · 2026
Clustering and predicting VCA patient populations - a machine learning approach into vascularized composite allotransplantation.
СтатьяНазвание 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.
Knoedler Leonard, Niederegger Tobias, Schaschinger Thomas, Stahl Carsten, Brandt Jule, Cetrulo Curtis L, Heiland Max, Hundeshagen Gabriel, Voss Jan O, Lellouch Alexandre G
Аннотация
BACKGROUND: Vascularized composite allotransplantation (VCA) enables functional and aesthetic restoration after devastating tissue loss. Yet, data-driven profiling of VCA recipients remains scarce. Identifying demographic and procedural patterns may improve transplant recipient selection, program planning, and perioperative care.
METHODS: VCA procedures in the Organ Procurement and Transplantation Network (OPTN, 1998-2023) were preprocessed, imputed, and clustered using weighted HDBSCAN. Resulting cluster patterns, temporal and geographic trends were analyzed. ARIMA models were applied to forecast patient demographics.
RESULTS: Among 107 recipients (mean age 37 ± 12 years; BMI 25 ± 5 kg/m²; 57% female), clustering analyses confirmed gender and VCA type as the primary differentiating dimensions, with face (19%) and upper-limb transplants (30%) predominantly performed in men and uterus transplants (31%) concentrated in United Network for Organ Sharing (UNOS) Regions 4 and 10. The facial VCA subgroup clustered regionally in Regions 1 and 9 (n = 13) and was characterized by comparatively high recipient age (42 ± 11 years) and long waiting times (246 ± 198 days). In the balanced clustering solution, 42% of recipients were classified as noise, highlighting substantial population heterogeneity. Temporal analyses showed declining activity in abdominal wall (11%) and upper-limb procedures, whereas ARIMA forecasts indicated stable age and BMI distributions through 2027.
CONCLUSIONS: This study provides the first machine-learning-based characterization of the national VCA recipient population using OPTN data. Rather than discovering entirely new recipient categories, the clustering framework successfully reproduced established population structures while identifying secondary regional and temporal patterns. Furthermore, density-based clustering highlighted substantial recipient heterogeneity through explicit outlier detection, thereby potentially providing a scalable framework for future registry-based analyses that incorporate immunological, outcome-related, and more granular clinical variables to enable earlier identification of atypical recipients and more targeted program planning.
METHODS: VCA procedures in the Organ Procurement and Transplantation Network (OPTN, 1998-2023) were preprocessed, imputed, and clustered using weighted HDBSCAN. Resulting cluster patterns, temporal and geographic trends were analyzed. ARIMA models were applied to forecast patient demographics.
RESULTS: Among 107 recipients (mean age 37 ± 12 years; BMI 25 ± 5 kg/m²; 57% female), clustering analyses confirmed gender and VCA type as the primary differentiating dimensions, with face (19%) and upper-limb transplants (30%) predominantly performed in men and uterus transplants (31%) concentrated in United Network for Organ Sharing (UNOS) Regions 4 and 10. The facial VCA subgroup clustered regionally in Regions 1 and 9 (n = 13) and was characterized by comparatively high recipient age (42 ± 11 years) and long waiting times (246 ± 198 days). In the balanced clustering solution, 42% of recipients were classified as noise, highlighting substantial population heterogeneity. Temporal analyses showed declining activity in abdominal wall (11%) and upper-limb procedures, whereas ARIMA forecasts indicated stable age and BMI distributions through 2027.
CONCLUSIONS: This study provides the first machine-learning-based characterization of the national VCA recipient population using OPTN data. Rather than discovering entirely new recipient categories, the clustering framework successfully reproduced established population structures while identifying secondary regional and temporal patterns. Furthermore, density-based clustering highlighted substantial recipient heterogeneity through explicit outlier detection, thereby potentially providing a scalable framework for future registry-based analyses that incorporate immunological, outcome-related, and more granular clinical variables to enable earlier identification of atypical recipients and more targeted program planning.