The personalized patient preference predictor proposed in 2024 has sparked widespread discussion. It mainly integrates machine learning with fine-tuning large language models to extract patients’ medical preferences by training and analyzing patients’ personalized sample data， to achieve better treatment. Combined with the current research situation， this paper pointed out its application dilemma： that is， the risks of inductive and analogical reasoning trigger its technical dilemma in accurate prediction， process explanation， and functional evaluation， ultimately leading to the ethical dilemma of the reduction of subject autonomy. Its solution paths were also clarified， including introducing substantive induction theory and substantive analogy theory to ensure the validity of sample data； implementing technology integration， perspective change， and evaluation system construction to improve the accuracy of prediction； as well as clarifying the priority of human subject status to protect the patient’s subjective autonomy. These showed that the personalized patient preference predictor has important social value and practical significance.

Objective:To investigate the types of non-tumor diseases in patients with cancer, and to explore the effects of those dis-eases on the diagnosis and treatment of cancer patients. Methods:We collected the medical records of cancer patients from January 2013 to December 2017 in Peking University Cancer Hospital, and screened for non-tumor diseases. The clinical records of the patients in this group were analyzed retrospectively, and the effects of those diseases on the diagnosis and treatment of tumors were dis-cussed. Results:Of the 1,323 cases of inter-hospital consultation, 1,153 cases of non-tumor disease (87.2％) were selected. There were 773 men (67.0％) and 380 women (33.0％) included. The median age was 62 (14-90) years. The primary tumor types included lung can-cer, gastric cancer, lymphoma, colorectal cancer, esophageal cancer, breast cancer, malignant melanoma, liver cancer, cholangiocarci-noma/gallbladder cancer, pancreatic cancer, and other tumors. Non-neoplastic diseases included cardiovascular disease in 356 cases (30.9％), respiratory system disease (17.0％) in 196 cases, digestive system disease in 107 cases (9.3％), skin and venereal diseases in 81 cases (7.0％), nervous system lesions (6.4％) in 74 cases, urinary system disease in 72 cases (6.2％), blood disease in 70 cases (6.1％), en-docrine and metabolic diseases in 47 cases (4.1％), autoimmune disease in 23 cases (2.0％), and other diseases (11.0％) in 127 cases. Impact on tumor diagnosis and treatment was as follows:direct, 771 cases (66.9％);no influence, 313 cases (27.1％);and uncertain, 69 cases (6.0％). Conclusions:Cardiovascular disease is a major non-tumor disease associated with cancer. Non-neoplastic diseases are important factors affecting the diagnosis and treatment plans of cancer.