In the late 20th century, programmed by computer engineers, AI was born based on a series of instructions (rules) created by humans, allowing technology to solve basic problems.
LTS: There are many industries affected by the influx of new technologies in the information age. With the impact of automation, computer science, and artificial intelligence (AI), entities such as doctors, hospitals, insurance companies and industries related to health care are not excluded. rules. But in healthcare, AI has a more positive impact than other industries.
First generation
It can be imagined that the way AI training is now similar to the approach of medical students, the AI system is also taught hundreds of algorithms to convert patient symptoms into diagnoses. This is considered the first generation of bringing health care rules into AI systems.
Decision algorithms are similar to how a tree grows, starting from the trunk (the patient's problem) and branching out from there. For example, if a patient complains of a severe cough, the doctor will first see if there is a fever. There will be 2 sets of questions depending on the case of fever/no fever. From the initial answer, it will lead to further questions about the condition. This in turn leads to further division. Ultimately, each branch is a diagnosis, which can range from bacterial, fungal or viral pneumonia to cancer, heart failure or dozens of other lung diseases.
In general, the first generation of AI can recognize problems but cannot analyze and classify medical records. As a result, early forms of artificial intelligence cannot be as accurate as doctors who combine medical science with their intuition and experience. And because of these limitations, rule-based AI is rarely used in clinical practice at other times.
Full automation
By the beginning of the 21st century, the second era of AI began with Artificial Narrow Intelligence (ANI), which means artificial intelligence that solves specific groups of tasks. The advent of neural networks that mimic the structure of the human brain has paved the way for deep learning technology. ANI operates very differently from its predecessor. Instead of providing rules predetermined by researchers, second-generation systems use huge data sets to distinguish patterns that would take humans a lot of time.
In one example, researchers fed thousands of mammograms into an ANI system, half of which showed malignant cancers and half showed benign cancers. The model can instantly identify dozens of differences in the size, density and shading of X-ray images, assigning each difference an impact factor that reflects the likelihood of malignancy. . Importantly, this type of AI does not rely on heuristics (some rule of thumb) like humans, but instead relies on subtle variations between malignant and normal tests that neither the radiologist nor the software designer knew.
Unlike rule-based AI, 2nd generation AI tools sometimes outperform doctors' intuition in terms of diagnostic accuracy. However, this form of artificial intelligence also shows serious limitations. First, each application has a specific task. That is, a system trained to read mammograms cannot interpret brain scans or chest X-rays. The biggest limitation of ANI is that the system is only as good as the data it has been trained on. A clear example of a weakness is when UnitedHealthcare relied on narrow AI to identify its weakest patients and provide them with additional medical services. When sifting through the data, the researchers later discovered the AI had made a disastrous assumption. Patients diagnosed as healthy only because their medical records received little medical care, while patients who used a lot of medical care services had low health ratings...
Future generations of AI will also allow people to diagnose diseases and plan treatments just like any doctor. Currently, the generative AI tool (Google's MED-PALM2) has passed the physician licensing exam with an expert-level score. Many other medical AI tools can now write the same diagnoses as doctors. However, these models currently still require physician supervision and do not have the ability to replace physicians. But at the current exponential growth rate, these applications are expected to become at least 30 times more powerful in the next five years. It is predicted that future generations of tools like ChatGPT will bring medical expertise into everyone's hands, fundamentally changing the relationship between doctors and patients.
VIET LE compiled
