Since the first artificial intelligence (AI)-enabled medical device received FDA approval for interpreting cervical slides in 1995, as of May 2023, there have been 521 FDA approvals for AI-enabled devices. Approval is provided.1 Many of these devices are for early cancer detection, an area of significant need as most cancers are diagnosed at a later stage. For most patients, early diagnosis means a higher likelihood of positive outcomes such as cure, less need for systemic therapy, and a higher likelihood of maintaining a good quality of life after cancer treatment. increase.
These extensive reviews are beyond the scope of a single article, but this article summarizes the main areas. AI and machine learning (ML) It is currently being used and studied for the early detection of cancer.
The first area is large database analysis to identify patients at risk for cancer or exhibiting early signs of cancer.these models Analyze electronic medical records, is a structured digital database that uses pattern recognition and natural language processing to identify patients with specific characteristics. These include individuals with signs and symptoms suggestive of cancer. People at risk of cancer based on known risk factors; or certain health measures related to cancer. For example, although the incidence of pancreatic cancer is relatively low, it remains the fourth leading cause of cancer death. Because of the low incidence, screening of the general population is neither practical nor cost-effective. ML can be used to analyze specific health outcomes such as new-onset hyperglycemia2 Specific health data from a questionnaire (3) to classify members of the population as at high risk for pancreatic cancer. This allows the screened population to be ‘enriched for pancreatic cancer’, thus increasing the screening yield and making it more cost-effective in the early stages.
Another area that leverages AI and ML learning is image analysis. Human vision is best in the center, representing her less than 3 degrees of vision. Peripheral vision has much lower special resolution and is better suited for fast movements and ‘big picture’ analysis. Furthermore, as demonstrated in a study that showed that even experts overlooked gorillas on CT when looking for pulmonary nodules, “inadvertent blindness” or critical Missing discoveries is one of human vulnerabilities.3 Machines are immune to fatigue, distraction, blind spots, or inadvertent blindness. In a study comparing deep learning algorithms to radiologists in the National Lung Screening trial, the algorithms performed better than radiologists in detecting lung cancer on chest x-rays.Four
AI algorithmic analysis of histological specimens serves as an initial screening tool and assistant as a real-time interactive interface during histological analysis.Five AI can diagnose cancer with high accuracy.6 Accurately determine the grade of prostate cancer, such as the Gleason score, and identify lymph node metastases.7 AI is also being explored in predicting genetic mutations from histological analyses. This can reduce costs and speed time to analysis. Both are limitations of today’s practice that limit universal genetic analysis of cancer patients.8 At the same time, however, it has also acquired a role in precision cancer therapy.9
Exciting and emerging fields where AI and deep learning are a combination of the above, such as combining large-scale data analysis with pathology assessment and/or image analysis. For example, using analysis of medical records and his CXR findings, Deep His Learning was used to identify patients at high risk for lung cancer and those who would benefit most from lung cancer screening. This has great potential. Especially since only 5% of his eligible patients for lung cancer screening are currently being screened.Ten
Finally, the holy grail of cancer detection: blood-based multiple cancer detection tests (many of which are already available and in development) often use AI algorithms to develop, analyze, and validate tests.11
It’s hard to imagine a healthcare sector without AI and ML impacts. AI is unlikely to replace doctors, at least not in the near future. It is used to improve physician performance and improve accuracy and efficiency. However, it’s essential to note that machine-human interactions are so complex that they’re just scratching the surface in this age. It is too early to assume that real-world results will resemble those seen in trials. Results involving human analysis and final decision making are subject to human performance. Human-machine interactions require training and research into human behavior to produce optimal results. For example, randomized controlled studies have shown increased polyp detection during colonoscopy using computer-assisted detection or AI-based image analysis.12 However, real data did not show similar results13 This is likely due to differences in the impact of AI on different endoscopists.
Artificial intelligence and machine learning are dramatically changing the way medicine is practiced, and cancer detection is no exception. Even in the world of medicine, which is usually slower to change than other fields, the pace of AI innovation is fast approaching.
