Citation

Sungheetha, Akey and R., Rajesh Sharma and Aroba, Oluwasegun Julius (2026) Explainable AI digital twin framework for early lung disease detection. Frontiers in Computer Science, 8. ISSN 2624-9898

Text fcomp-8-1652980.pdf - Published Version Restricted to Repository staff only Download (992kB)

Abstract

Introduction: Digital twin technology creates virtual replicas of physical systems, enabling real-time monitoring and predictive analytics through continuous data synchronization. This study presents an explainable artificial intelligenceenhanced digital twin framework specifically designed for the early detection of chronic lung abnormalities in urban young adults aged 20–35 years. Methods: Analysis of 4,247 patients from the Delhi metropolitan area revealed a 29.3% prevalence of structural lung damage, including bronchiectasis, emphysema, and fibrosis. The framework integrates multimodal physiological sensors, environmental pollution monitoring, and lifestyle data through advanced fusion algorithms. Mathematical modeling incorporates bronchial resistance Rb = 2.34 ± 0.45 cmH2O/L/s, lung compliance CL = 0.187 ± 0.032 L/cmH2O, and deterioration rate λdet = 0.0156 ± 0.0023 per month from longitudinal monitoring. Blockchain integration ensures data security with hash validation efficiency ηhash = 0.987 and real-time processing latency τresp = 127.3 ± 15.7 ms. Environmental factor integration, including the air quality index AQI = 247 ± 67, enables personalized risk stratification accuracy βrisk = 0.876 ± 0.045. Results: Core performance metrics demonstrate explainability coefficient ξexp = 0.847 ± 0.023, prediction accuracy αpred = 0.923 ± 0.034, and early detection capability extending tearly = 6.7 ± 1.2 months before clinical symptoms. Validation across 1,847 test subjects achieved sensitivity, Searly = 0.891, specificity, Spearly = 0.876, and positive predictive value (PPV) = 0.834. Environmental factor integration, including the air quality index AQI = 247 ± 67, enables personalized risk stratification accuracy βrisk = 0.876 ± 0.045. Statistical analysis confirmed significant improvements in diagnostic timing (p < 0.001), intervention effectiveness (p < 0.001), and patient outcomes compared to conventional approaches. Discussion: Clinical implementation demonstrates 68.4% reduction in diagnostic delays, 73.6% improvement in intervention timing, and annual healthcare cost savings of �C = $2, 847 per patient.

Item Type:	Article
Uncontrolled Keywords:	Blockchain security, digital twin
Subjects:	Q Science > QA Mathematics > QA71-90 Instruments and machines > QA75.5-76.95 Electronic computers. Computer science
Divisions:	Faculty of Information Science and Technology (FIST)
Depositing User:	Ms Rosnani Abd Wahab
Date Deposited:	05 May 2026 07:36
Last Modified:	08 May 2026 06:30
URII:	http://shdl.mmu.edu.my/id/eprint/15893

Downloads

Edit (login required)