A practical guide to the implementation of artificial intelligence in orthopaedic research-Part 2: A technical introduction

Abstract

Recent advances in artificial intelligence (AI) present a broad range of possibilities in medical research. However, orthopaedic researchers aiming to participate in research projects implementing AI-based techniques require a sound understanding of the technical fundamentals of this rapidly developing field. Initial sections of this technical primer provide an overview of the general and the more detailed taxonomy of AI methods. Researchers are presented with the technical basics of the most frequently performed machine learning (ML) tasks, such as classification, regression, clustering and dimensionality reduction. Additionally, the spectrum of supervision in ML including the domains of supervised, unsupervised, semisupervised and self-supervised learning will be explored. Recent advances in neural networks (NNs) and deep learning (DL) architectures have rendered them essential tools for the analysis of complex medical data, which warrants a rudimentary technical introduction to orthopaedic researchers. Furthermore, the capability of natural language processing (NLP) to interpret patterns in human language is discussed and may offer several potential applications in medical text classification, patient sentiment analysis and clinical decision support. The technical discussion concludes with the transformative potential of generative AI and large language models (LLMs) on AI research. Consequently, this second article of the series aims to equip orthopaedic researchers with the fundamental technical knowledge required to engage in interdisciplinary collaboration in AI-driven orthopaedic research.

Level of evidence: Level IV.

Keywords: artificial intelligence; machine learning; orthopaedics; research methods; sports medicine.

Figure 1

The diagram illustrates the subdomains of narrow artificial intelligence (AI), including levels of supervision and the most frequently applied methods according to each subdomain.

Figure 2

A schematic representation of commonly performed machine learning tasks. (a) Regression: a line (yellow) providing the best fit to the data (blue dots) is applied and the model can be used to predict a continuous outcome (y) based on one or several predictor variables (x). (b) Dimensionality reduction: enables a reduction in the number of variables considered for modeling an outcome through feature selection and/or extraction. This is illustrated by reducing a three‐dimensional data set (blue dots) into two principal components (yellow lines: PC1 and PC2) through principal component analysis (PCA). (c) Classification methods are used to assign data points (blue dots) into two or more classes (yellow and blue triangles) based on differences in characteristics, which the model can interpret as boundaries to separate data. (d) Clustering involves the separation of input data into two or more clusters based on similarities and differences in a set of characteristics. The illustration displays three patient subgroups (yellow, blue and purple ovals) identified within a hypothetical data set (blue dots) using a clustering approach. (e) Neural networks are organised in layers of algorithms that mimic the interconnectedness of neurons in the brain. The illustration displays a neural network with interconnected nodes arranged in multiple connected layers of a certain depth. Data at the input level (dark blue node) are transmitted through subsequent layers of the network (light blue nodes) until the layer providing the output (yellow nodes) is reached.

Figure 3

The diagram displays the basic components of predictive artificial intelligence (AI) models, including labelled and unlabelled data at the input level (x) and numeric, discrete, probability‐based or class‐based variables at the output level (y).

Figure 4

The diagram displays the basic components of generative artificial intelligence (AI) models, which accept structured or unstructured data as input (x) and return text, images, audio, video or other generated content as output (y).