Artificial Intelligence is already used widely in the field of manufacturing, where deep learning algorithms are being used for the maintenance of machines in production processes and for predicting anomalies. So why not use AI to predict problems in the various complex systems of the human body? AI in the field of medicine is still at a nascent stage. With the use of AI, scientists are developing new ways of detecting diseases with incredible accuracy. So much so, that in the future we could have robots replacing certain critical tasks done by humans!
Paras Lakhani and Bhaskar Sundaram at the Radiological Society of North America have been successful in detecting the presence of tuberculosis (TB) by employing deep learning algorithms to study patient’s chest x-ray images. The researchers first trained the AI with data from hundreds of x-rays of patients with and without TB. At the end of the training, they tested the AI with 150 new x-rays. They found that the AI had a 96% accuracy rate, which is far greater than the average rate of a human radiologist. However, the application of this particular AI algorithm does not extend beyond TB because it was not designed for other diagnoses. What is truly significant is that this approach, when coupled with a radiologist’s opinion, improved diagnostic accuracy to almost 100%! In countries where trained radiologists are in short supply, AI programs such as this will be very useful.
Research into neural networks and pathology at Google Brain, a deep learning artificial intelligence research team at Google, revealed some interesting facts. Using microscopic specimen images, the team successfully trained an AI bot to detect with high accuracy the spread of breast cancer in lymph node tissues. Isolating cancer cells on a specimen slide requires a series of procedures that are time-consuming and laborious, in addition to being prone to human error. However, the AI bot was able to process large swathes of imaging with significant accuracy to find cancerous anomalies.
Deep learning is assisting medical professionals and researchers to serve the healthcare industry better. With over 300 million diagnostic radiology images being taken in the US alone, there is pressure on the healthcare industry to operate more efficiently and accurately. With regards to incoming medical cases, Enlitic’s deep learning technology is able to interpret medical images in a matter of milliseconds to determine the priority of the case. The deep learning method is also being implemented in the field of genomics to help predict the kind of diseases that can affect a patient. You can now monitor your children’s health using the Oto, a smart otoscope that allows you to capture a video of the eardrum from your iPhone. Another example of deep learning, cellscope uses data to track your children’s health and book appointments with doctors precisely when needed.
From drug discovery to medical imaging, to Alzheimer’s detection and genomics to even insurance fraud, AI has applications in a wide range of fields. For now, the main use of AI is to decrease diagnosing times and increase efficiency and accuracy. This in turn will help doctors provide better medical treatment for their patients.