Source – https://www.forbes.com/

It was no surprise that one of the first applications of neural networks, a core modern technique of artificial intelligence (AI), was in radiology. Vision applications were pointed at the task of looking at x-ray and CT scan images in order to identify cancerous tumors. That helped drive the rebirth of a focus on AI. The evolution of AI and machine learning (ML) has begun to help the medical community move past pure identification of disease into more predictive and prescriptive arenas of healthcare. One example is in the area of lung disease.

The initial identification of tumors by AI was an important step in improving the accuracy of both diagnosis and neural networks. The next obvious step in healthcare was to identify other diseases. With the pulmonary system, the lungs in particular, identifying cancer extended to recognizing issues brought on by emphysema. One of the things I learned about that disease, in the course of investigation, is that damaged sections of the lungs don’t just stop working. When we breath, we take in gases and then the lungs transfer gases such as oxygen into the blood stream. One effect of emphysema is that the failure to transfer the gas means that air is trapped in parts of the lung, they inflate, and that prevents air from moving to other parts of the lung that are still healthy.

The identification of the biomarkers of emphysema, and other diseases, can be identified with neural networks. However, note my usual soapbox’s that AI is a tool and that ML is now more than the pure AI definition. It’s one thing to identify a problem. It is beyond that to use the knowledge to plan and perform actions. “Neural networks are critical to identifying issues in images,” says Susan Wood, PhD, President & CEO, VIDA. “However, it’s not just about AI finding things in images or doing it faster, but in improving the holistic impact on care pathways and patient treatment.”