| Chronic obstructive pulmonary disease(COPD)is a serious chronic respiratory disease,which seriously threatens the health of Chinese people.The quality of life of COPD patients is much lower than that of healthy people,and the disease causes a huge economic burden to patients and their families.What is not commensurate with the high prevalence,mortality and disease burden of COPD in China is the insufficient capacity of diagnosis and treatment of COPD,the low awareness rate of COPD and the low prevalence rate of pulmonary function examination in Chinese primary medical institutions.Strengthening capacity building at the primary care level is an important measure to reduce the prevalence and disease burden of COPD.Enhancement of grassroots capacity should cover all aspects of screening,diagnosis and standardized management of COPD.How to improve the ability of screening and disease risk assessment of COPD high-risk groups at the grassroots level,the ability of quality control of spirometry examination,and the ability of standardized management of COPD patients,so as to bridge the significant gap between the best evidence-based medicine practice and practical application.In this study,the screening capacity of COPD high-risk population at the grassroots level was improved through the development and application of the screening model for COPD high-risk population.Through the development and application of spirometry examination quality control model to improve the ability of standardized spirometry examination at the grass-roots level.Through the development and application of the clinical decision aid system of COPD composed of the management modules of COPD stable phase and acute exacerbation phase,the standard management ability of COPD at the primary level will be improved,and the pain points in the current management of COPD at the primary level will be solved.In this study,a convenient and effective Clinical Decision Support System(CDSS)was developed to assist the whole process from COPD screening to diagnosis and standardized management at the primary level,and to provide high-quality medical care for patients by coupling with different types of primary health care systems.The CDSS for COPD consists of three parts:screening and risk assessment system for COPD high-risk groups,quality control system for spirometry examination,and clinical decision support system for COPD.In this study,18 different machine learning algorithms which are popular in recent literatures were selected to develop a screening model for COPD high-risk population based on the data of 1875 residents’ electronic health records.We found that among the COPD screening tasks based on public health data,Catboost was the best model for comprehensive evaluation,with accuracy of 99.25%,and sensitivity of 94.81,nearly 6 percentage points higher than the baseline model after optimization,AUC of 0.9985.It is worth noting that the precision is 1,indicating that the non-high-risk population is not diagnosed as high-risk population,and the misdiagnosis rate is 0.The comprehensive evaluation result of Stacking model is close to that of CatBoost,and the overall generalization ability of the integrated learning model is better than that of the single model by observing the test results of external data.In the process of external verification,the spirometry examination quality control system indicated that 39.29%of the lung function tests could meet the quality control standards of spirometry examination tests after improvement,and the compliance rate of quality control rules increased by 28.44%.This study constructed the knowledge map according clinical COPD management guidelines.Through the workflow model to achieve the whole process of decision support of COPD diagnosis,illness evaluation,differential diagnosis,referral advice evaluation,treatment and follow-up management,helping grassroots doctors to management of patients with COPD according to the guidelines. |
PDF Full Text Request