Prediction method for tumor diagnosis based on medical big data

Abstract

The invention discloses a prediction method for tumor diagnosis based on medical big data. The method comprises the steps of data mining and model construction. The data mining comprises early-stage collection of tumor marker big data in recent three years, the model construction comprises 5-8 tumor benign and malignant models, logistic regression is adopted. Related data are analyzed and predicted through four machine learning methods including a random forest, a support vector machine and reinforcement learning. The method is reasonable, researchers will collect markers in various medical records related to physical examination and tumors, the markers comprise indexes such as age and sex. Whether the tumor is malignant or benign is judged according to known tumor data, the iterative training is carried out, and the benign or malignant tumor is estimated through various algorithms of machine learning, so that a potential prediction tool can be provided for the cancer field. The method for distinguishing benign and malignant tumors through machine learning has very beneficial application prospects, and is reliable in result, high in accuracy and good in specificity.

Description

technical field [0001] The invention relates to the technical field of medical big data utilization, in particular to a prediction method for tumor diagnosis based on medical big data. Background technique [0002] Cancer is a complex disease, and its incidence is increasing year by year. The treatment of cancer is also an astronomical expenditure in various countries. How to diagnose cancer early and treat it in time is a subject that scientists from all over the world have been studying unremittingly. How to expand the screening scale and reduce medical costs is also a huge challenge. Each tumor often has multiple driver genes and complex signal transduction pathway abnormalities, but early diagnosis of cancer has always been a thorny problem. How to use simple means to It is an important and urgent problem to predict early-stage tumors or judge through the blood test indicators of patients and through algorithms. To the above problems, a solution is proposed below. Con...

AI Technical Summary

Problems solved by technology

[0002] Cancer is a complex disease, and its incidence is increasing year by year. The treatment of cancer is also an astronomical expenditure in various countries. How to diagnose cancer early and treat it in time is a subject that scientists from all over the world have been studying unremittingly. How to expand the screening scale and reduce medical costs is also a huge challenge. Each tumor often has multiple driver genes and complex signal transduction pathway abnormalities, but early diagnosis of cancer has always been a thorny problem. How to use simple means to It is an important and urgent problem to predict early-stage tumors or judge through the blood test indicators of patients and through algorithms

Images