AI-Based Heart Failure Detection

AI-Based Heart Failure Detection

AI-Based Heart Failure Detection

An accurate prognosis, whether good or bad, can help patients develop good coping skills. But there are many diseases that are quite tricky to prognose. In fact, the prognosis is always much more difficult than diagnosis in any case. Also, the monitoring and prognosis of cardiovascular conditions usually require the use of expensive, sophisticated equipment and intrusive procedures.

However, advancements in disease diagnostics with AI are giving rise to simple, high-precision AI-based monitoring and prognostic equipment for mainstream use. Such equipment, which are poised to make better predictions than doctors, are designed to extrapolate patterns from vast collections of patient data and to use those patterns to identify early signs of heart diseases in patients.

Key Technologies Exploring AI to Detect Heart Disease

There have been a handful of AI-based inventions that detect heart conditions early enough. One of the most promising among these is a product of studies carried out by Dr. Declan O’Regan and his colleagues at the Imperial College London. The project revolves around using machine learning, a subset of AI, to prognose diseases with even greater accuracy than doctors.

In this system, the computer is trained to make predictions by sorting through the data of hundreds or even thousands of patients using preset algorithms. The computer is able to make better predictions than doctors because it identifies exquisite details of the accumulated data which the human senses cannot perceive.

Another similar development that uses AI to detect diseases, of which details have been published in the Journal of the American Medical Information Association (JAMIA), makes predictions by tracking the temporality in a data set. It uses deep learning, which is the capacity to identify key features in a data set, to diagnose heart disease with AI.

The model uses a recurrent neural network (RNN) to track temporal relations among elements of data sets of patients. With deep learning, the model is able to identify the sequences of events that lead up to the various stages of a heart condition.

Machine Learning to Diagnose Heart Disease with AI

Another similar model has been developed by researchers at the USC Viterbi School of Engineering. In this model, machine learning is combined with pulse data to detect key risk factors for cardiovascular diseases using just a smartphone. The risk factor which it’s keen to identify is arterial stiffness — the weakening of elasticity and the increase in rigidity in arteries, known to be a precursor to cardiovascular diseases, diabetes, renal failure, and many other diseases.

Pulse Wave using the Phone's Camera

Pulse Wave using the Phone’s Camera

The pulse energy flowing from a stiff aorta to small vessels can damage organs. Arterial stiffness can be identified through the measurement of pulse wave velocity. The current technologies for measuring pulse wave velocity include MRI, tonometry, and electrocardiogram, which all come with various drawbacks including high cost and very complex operations.

With this new technology for identifying arterial stiffness, anyone can detect arterial stiffness through a smartphone app that uses a smartphone camera to record pulse waves underneath the skin.

A research team from Nanyang Technological University and Tan Tock Seng Hospital, both in Singapore, takes AI-based heart failure detection to cloud facilities through connected devices. This model integrates a stethoscope-like acoustic sensor and cloud-based AI software in smartphones to track the presence of excess fluids in the lungs — a sign of an impending heart failure. This model, which detects the warning signs of heart failure within 10s after being placed on a patient’s chest or back, has proved to prognose heart failure with over 92% accuracy.

The Advancements in the Pipelines for Disease Diagnostics with AI

Dr. O’Regan and his colleagues are looking to further expand the capabilities of their model of AI for diagnosing heart diseases like dilated cardiomyopathy — a gateway to heart failure, pulmonary hypertension –a rare condition in which the right side of the heart is damaged, and even other ailments like a stroke.

The tests for the prognostic accuracy of these models, which were implemented by processing data through the AI algorithm to make predictions about the various stages of heart disease cases that had already unfolded, showed the models can have prognosis with over 92% accuracy. The team is now set to roll out the use of the models for the prognosis of live clinical patients. The prognosis will help direct the monitoring efforts of those undertaking the patient’s care.

Conclusion

AI and machine learning are increasingly being used to provide solutions in various aspects of health care. And each successful development serves as a springboard for another. The advancements in the application of AI technology in the area of medical prognosis has been supercharged by the exploration of deep learning principles.

Leave a Comment

Your email address will not be published.