Project Details
Description
The cancer medical expenses in 2013 were NTD 74.8 billion, but in 2017, it increased to NTD 93.3 billion, accounting for 13% of the total health insurance 700 billion, and continued to increase. It is well known that prevention is better than cure, but most of the current resources (money) are invested in medical diagnosis and treatment. This type of expenditure structure may cause cancer treatment to be an unaffordable future. The reasons for negligent prevention include (1) no obvious goals, (2) slow speed in improvement, and (3) the patients feel no pain for the disease, and the most important is the lack of accurate risk prediction.
Using breast cancer and lung cancer as examples, the positive predictive rates (PPV) of mammography and low-dose CT are 5.0% and 6.4%, respectively, and high false positives rare results in poor effectiveness. Prevention is an urgent medical task, and it must be precise, timely, personalized and enforceable. Therefore, this plan provides a simpler, cheaper, and 10 times more effective technology. This overcomes the calculation method of 10 or 20 variables in the previous statistical method, and performs multidimensional variable data calculation based on the imematrix, personalized diseases and medications. We aim to
develope the biomarker for cancer screening based on artificial intelligence, which does not require any invasive examination, only needs medical history data (diagnosis + medication) in the past 3 years from the patients, that is, can predict more than 10 kinds in the next 12 months. Moreover, the prediction accuracy (AUROC)
of cancer risk is over 85%.
Taking 10,000 breast cancers as an example, the average personal health insurance expenditure (NTD 66,000) of patients with early detection of breast cancer is onequarter of that of advanced patients (NTD 158,000), and it can save at least 100 million NT dollars in commercial insurance claims expenses. Therefore, the early awareness management of major diseases of AI in this project will be able to increase the number of stage 0 and stage 1 cancers,
detect cancer precursors early, and prevent cancer progression.
Using breast cancer and lung cancer as examples, the positive predictive rates (PPV) of mammography and low-dose CT are 5.0% and 6.4%, respectively, and high false positives rare results in poor effectiveness. Prevention is an urgent medical task, and it must be precise, timely, personalized and enforceable. Therefore, this plan provides a simpler, cheaper, and 10 times more effective technology. This overcomes the calculation method of 10 or 20 variables in the previous statistical method, and performs multidimensional variable data calculation based on the imematrix, personalized diseases and medications. We aim to
develope the biomarker for cancer screening based on artificial intelligence, which does not require any invasive examination, only needs medical history data (diagnosis + medication) in the past 3 years from the patients, that is, can predict more than 10 kinds in the next 12 months. Moreover, the prediction accuracy (AUROC)
of cancer risk is over 85%.
Taking 10,000 breast cancers as an example, the average personal health insurance expenditure (NTD 66,000) of patients with early detection of breast cancer is onequarter of that of advanced patients (NTD 158,000), and it can save at least 100 million NT dollars in commercial insurance claims expenses. Therefore, the early awareness management of major diseases of AI in this project will be able to increase the number of stage 0 and stage 1 cancers,
detect cancer precursors early, and prevent cancer progression.
| Status | Finished |
|---|---|
| Effective start/end date | 7/1/20 → 12/31/20 |
Keywords
- AI
- time matrix
- earlier cancer prediction
- health checkup center
- commercial insurance
Fingerprint
Explore the research topics touched on by this project. These labels are generated based on the underlying awards/grants. Together they form a unique fingerprint.