AI’s potential has brought about a paradigm shift in the healthcare and research industries. This week’s guide discusses its growing developments in the same.
AI Aids Eye Testing
Artificial Intelligence (AI) has a long list of applications in the healthcare industry. The newest addition to this list is a hi-tech screening tool recently developed by Google and a team of international researchers. It detects Diabetic Retinopathy — a diabetic complication in the eye.
A recent study conducted was with this AI-powered tool at two eye care centres in India. These are at Aravind Eye Hospital, Madurai and at Sankara Nethralaya, Chennai. They screened over 3,000 patients with diabetes, and it was seen that the AI’s performance increasingly exceeded conventionally used manual grading methods used in the identification of diabetic retinopathy. The AI-powered tool had a specificity and sensitivity of 90% as per the results published in JAMA Ophthalmology.
At first, a specialised retinal fundus camera is used to take photos of the eye. Once the required images are captured, it is then fed into the computer containing the AI-powered tool. It then screens the images for Diabetic Retinopathy and gives an instant report of the patient along with the recommendations. According to the International Clinical Diabetic Retinopathy scale, the AI-powered tool has been taught to grade the severity of the condition – from none, mild, moderate, severe to proliferative, making the process easy to detect and understand.
AI Identifies Optimal Material Formula
Nanostructural layers posses innumerable potential properties. However, identification of the most suitable ones most often require long-term and expensive experiments. In order to find the optimal parameters for applications of nanostructured layers, it is necessary to manually conduct countless experiments under different conditions with different compositions – making it an incredibly complex and prolonged process. The findings yielded by these experiments are known as Structure Zone diagrams, from which the parameters are read and inferred.
However, a PhD researcher at Lars Banko, along with a team has ventured a shortcut for this process, using a machine learning algorithm. They modified a generative Machine Learning model – which is capable of reliably predicting the properties of a Nanostructured layer. The team trained an algorithm to generate images of the surface of a thoroughly researched model layer of aluminium, chromium and nitrogen using specific parameters, to predict the layer properties under respective conditions. The results were conclusive, considering that the researchers were able to compare the results of the calculations with that of the experiments and analyse the reliability of its prediction.
Are you curious to learn more about AI? Explore careers and upskill with Great Learning’s Artificial Intelligence and Machine Learning program.
