Construction and verification of prognostic classification model for elderly cancer patients in rural areas based on machine learning algorithm
10.3760/cma.j.cn211501-20231231-01449
- VernacularTitle:基于机器学习算法的农村老年癌症患者预后分类模型的构建与验证
- Author:
Li CHANG
1
;
Zhihua YANG
;
Jiang ZHAO
;
Qin YUE
;
Honghong SHEN
;
Chunxiu FAN
;
Juan XIE
Author Information
1. 陕西省肿瘤医院护理部，西安 710061
- Keywords:
Rural area;
Aged;
Cancer;
Prognosis;
Machine learning
- From:
Chinese Journal of Practical Nursing
2024;40(21):1661-1670
- CountryChina
- Language:Chinese
-
Abstract:
Objective:The classification model of prognosis of elderly cancer patients in rural areas of Shaanxi province was constructed based on machine learning algorithm, and its effectiveness was verified, providing reference for early prognosis recognition and intervention treatment.Methods:Using a multicenter, cross-sectional survey method and convenience sampling method, 597 elderly cancer patients in rural areas hospitalized in 9 different medical institutions (Shaanxi Cancer Hospital and its member units of specialty alliance) in Shaanxi Province from July to August 2022 were selected as the research objects, and a variable database of "basic information", "self-care ability", "symptoms", "comprehensive needs" and "quality of life" of elderly cancer patients in rural areas was established. Machine learning and statistical analysis were carried out to explore the important prognostic characteristics of elderly cancer patients in rural areas, and a prognostic classification model for elderly cancer patients in rural areas was constructed and verified.Results:The 597 valid questionnaires were ultimately collected.Among the 597 elderly cancer patients in rural areas, 207 were males and 390 were females, aged (69.56 ± 8.84) years. The results of cluster exploration showed that the prognosis of elderly cancer patients in rural areas was divided into three categories: good, medium and poor. The areas under the working characteristic curves of the subjects in good, medium and poor were 0.84, 0.79 and 0.69, respectively. The results of characteristic exploration showed that 10 indicators of "whether accompanied by metastasis", "distress", "sadness", "numbness", "eating", "walking", "fatigue", "forgetfulness", "fun of life" and "relationship with others" were important characteristic indicators of prognosis evaluation of elderly cancer patients in rural areas. There were statistically significant differences among the 10 important characteristic indicators in the three categories of prognosis of rural elderly cancer patients ( χ2=21.07, H values were 18.51-144.38, all P<0.01). There were statistically significant differences in the three categories of "comprehensive needs", "quality of life", "self-care ability" and "symptoms" ( H values were 519.40, 40.80, 103.69, all P<0.01). Conclusions:The construction and verification of a prognostic classification model for elderly cancer patients in rural areas based on machine learning algorithm can effectively explore the important characteristic indicators and prognostic classification of elderly cancer patients in rural areas, and provide basis and guidance for clinical medical staff to make individualized plans.
