Artificial Intelligence and Machine Learning in Pathology: The Present Landscape of Supervised Methods.

Area of Science:

• Health Informatics
• Medical Artificial Intelligence
• Computational Pathology

Background:

• Growing interest in AI and ML applications within healthcare.
• Need for cross-disciplinary understanding among pathologists and laboratorians regarding AI/ML.
• AI/ML tools are being developed to impact medical practice, including pathology.

Purpose of the Study:

• To provide foundational knowledge of machine learning categories.
• To introduce the bias-variance trade-off in supervised learning.
• To describe common supervised machine learning algorithms for medical applications.

Main Methods:

• Review of machine learning categories: supervised, unsupervised, and reinforcement learning.
• Explanation of the bias-variance trade-off concept.
• Overview of supervised learning study design and common algorithms.

Main Results:

• Definitions and explanations of supervised, unsupervised, and reinforcement learning.
• Introduction to the bias-variance trade-off.
• Description of algorithms including linear regression, logistic regression, Naive Bayes, k-nearest neighbor, support vector machine, random forest, and convolutional neural networks.

Conclusions:

• Cross-disciplinary literacy is crucial for optimal AI/ML tool design in healthcare.
• Understanding ML fundamentals empowers pathologists and laboratorians to leverage AI.
• This review serves as a guide to essential ML concepts for medical professionals.