AI: ML and DL

As AI is becoming increasingly useful in the healthcare environment, methods such as ML (machine learning) and DL (deep learning) are being widely used in the “prediction and diagnosis of several diseases, especially those whose diagnosis is based on imaging or signaling analysis.”

ML can be split into three sections based on learning methods: supervised, unsupervised, and reinforced learning. Additionally, a few steps can be used to describe ML. Processing data is the first priority, fundamental for improved quality of data and a reduction of incorrect results. Then the ML model chosen for that image classification will have crucial features implemented to aid (and sometimes perform) the job of physicians. ML algorithms guide in detecting patterns in medical data that would have completely missed human eyes. Additionally, ML is used in diagnostic support systems, “risk assessment tools, and patient monitoring applications.” However, ML requires experts to identify some relevant features prior to models. This makes ML dependent on human operation, limiting the possible future ML could have. This is where DL steps in. DL learns autonomously from raw data and has the ability to analyze more complex data for medical imaging and genomics. Using neural networks with multiple layers, DL has been incorporated into technology such as X-rays and MRI scans. The artificial neural network (ANN) is modeled after human brain cells, where multiple processing components blend to form a more advanced and intricate system. The small components form layers in every algorithm, and in the synthetic type, similar to the human brain, the components are “replaced by a sum and an activation function.”

DL is equipped to function to detect cancerous lesions and foreshadow heart-related risks. One of the first applications of DL was for image processing, specifically Magnetic Resonance Imaging (MRI) scans. The number of layers DL contains benefits it in terms of integrating diverse data and being able to provide generalization when the focus is not just targeted on accuracy. The possibility of more wearable devices that monitor a patient’s medical condition is increasing due to ML and DL. This is especially important because these devices facilitate managing diseases and grant patients access to their analytics quickly in real-time.

References:

Ghaffar Nia N, Kaplanoglu E, Nasab A. Evaluation of artificial intelligence techniques in disease diagnosis and prediction. Discover Artificial Intelligence [Internet]. 2023 Jan 30;3(1). Available from: https://www.ncbi.nlm.nih.gov/pmc/articles/PMC9885935/ 

Flam S. Machine Learning in Healthcare - Benefits & Use Cases [Internet]. ForeSee Medical. 2020. Available from: https://www.foreseemed.com/blog/machine-learning-in-healthcare 

Miotto R, Wang F, Wang S, Jiang X, Dudley JT. Deep Learning for healthcare: review, Opportunities and Challenges. Briefings in Bioinformatics. 2018;19(6):1236–46. 

Roy R. Understanding the difference between AI, ML and DL!! [Internet]. Medium. 2020. Available from: https://towardsdatascience.com/understanding-the-difference-between-ai-ml-and-dl-cceb63252a6c