Lisheng Huang* Department of Radiotherapy, Cancer Hospital of Shantou University Medical College, Shantou 515041, Guangdong Province, China
Wenzhao Lin Department of Medical Oncology, Shantou Central Hospital, Shantou 515041, Guangdong Province, China
Shuhan Yu Department of Medical Oncology, Shantou Central Hospital, Shantou 515041, Guangdong Province, China
Objective: This study aims to explore application value of artificial intelligence in identification of nature of lung cancer patients in China, analyze its existing problems and advantages, and provide a basis for future solutions. Methods: We randomly selected lung cancer patients (N=121) who received antitumor therapy in our hospital from October 2017 to October 2021. 100 of them met inclusion criteria of WFO. The diagnosis and treatment plan of WFO is divided into three categories, namely: "recommended", "considerable" and "not recommended". When clinical oncologists choose "recommended" or "considerable" regimens, they are considered consistent, and rest are considered inconsistent. Descriptive statistics were performed on case characteristics of all patients using Microsoft Excel, and patients were grouped according to their case classification, gender, age, and whether or not to undergo surgery, and their consistency was analyzed. SPSS17.0 software was used for statistical analysis, and logistic regression model was used to estimate probability ratio and 95% confidence interval of above factors. Result: Among all cases, 21 cases were not applicable to WFO. Among 100 applicable cases, median patient age was 61 years, 70% were male and 30% were female. Among them, 21% had surgery, and 79% had no surgery. Small cell lung cancer patients accounted for 19%, and non-small cell lung cancer patients accounted for 71%. Among all enrolled cases, 85% of diagnosis and treatment plans proposed by WFO were consistent with those of doctors. The consistency of patients with small cell lung cancer (SCLC) was 89.48%; consistency of patients with non-small cell lung cancer (NSCLC) was 83.96%. According to tumor stage, 83.33% of patients with stage II, 83.33% of patients with stage III and 85.94% of patients with stage IV had same diagnosis and treatment plan. According to gender classification, consistency between diagnosis and treatment plan proposed by WFO and that of doctors is 88.57% for male patients and 76.67% for female patients. According to classification of operation or not, consistency between diagnosis and treatment plan proposed by WFO and that of doctors was compared, consistency of patients after surgery was 85.72%, and consistency of patients without surgery was 84.82%. According to age classification, according to diagnosis and treatment plan proposed by WFO, 87.93% of patients aged 60 years or older are consistent with doctors, and 80.95% of patients younger than 60 years old are consistent with our center. According to classification of squamous cell carcinoma and adenocarcinoma, squamous cell carcinoma accounts for 28.40% and adenocarcinoma accounts for 71.60%. The diagnosis and treatment plan proposed by WFO was consistent with doctors in 86.95% of squamous cell carcinoma patients and 82.76% of adenocarcinoma patients. Adenocarcinomas were divided into EGFR mutation, EGFR wild-type and undetected according to gene mutation. The consistency of EGFR mutation was 73.34%, consistency of EGFR wild was 85.71%, and consistency of without genetic testing was 86.66%.Small cell lung cancer is divided into limited stage and extensive stage. 77.78% of patients with limited stage according to WFO's diagnosis and treatment plan are consistent with doctors, while rate of extensive stage is as high as 100%. According to logistic regression model analysis, factors such as lung cancer stage, histological type, gender and age had no effect on consistency. Conclusion: The diagnosis and treatment plan proposed by WFO for enrolled cases is highly consistent with that of clinical oncologists;2. The difference in selection of lung cancer diagnosis and treatment plans between artificial intelligence and clinical oncologists may be affected by factors such as patient constitution, drug availability, and medical insurance plans in East and West;3. If artificial intelligence wants to be applied quickly and comprehensively in China, it is necessary to accelerate its localization and improve its consistency.
Keywords:Artificial intelligence, lung cancer, identification, consistency