30 March 2023

---

Artificial intelligence and deep learning

Lara Figini

---

Artificial intelligence (AI) refers to computer programs capable of performing tasks that typically require human cognition. AI programs were initially created as expert systems, mimicking human decision-making using conditional logical statements. Machine learning is a subtype of artificial intelligence in which machines learn to perform tasks by learning patterns in data rather than by being programmed with explicit rules and statements.

Deep learning is a particular form of machine learning based on artificial neural networks (ANN) inspired by the human nervous system. The term "deep" refers to the network architecture of the multi-level model. In traditional machine learning, the model learns features of the data from manual extraction, while in deep learning, representations (including feature extraction) are learned from the raw input data.

Deep learning has been found to outperform conventional machine learning when dealing with complex and computationally intensive tasks such as image analysis and natural language processing. In recent years machine and deep learning have made significant progress and are increasingly used in consumer products such as smartphones, cameras, and web search engines.

In parallel, the digitization of healthcare, including the generation of digital health data such as electronic records or images, has facilitated the growing application of machine and deep learning in medicine and dentistry. Both clinicians and dental researchers have the task of becoming familiar with these technologies, given their increasing relevance in the dental area.

Materials and methods

In a recent review to be published in the Journal of Dentistry, the authors explained the basics of the concept of deep learning by explaining commonly used terms and describing different deep learning approaches, their methods, and results.

This review is based on the collection of the latest studies, medical primers, and the state of the art of research on artificial intelligence and deep learning. Deep learning in healthcare and dentistry Artificial neural networks (ANNs) were first introduced in 1943 by Warren McCulloch and Walter Pitts, who presented a simplified computer model of how biological neurons in animal/human brains could perform complex computations.

However, computer hardware restrictions and software application difficulties severely limited ANN's performance for decades. After that, expert systems have been developed, such as the MYCIN for clinical diagnostics (which can assist doctors in diagnosing bacterial infections in patients and in selecting the most appropriate antibiotics) and such as the "Oral Radiograph Diagnosis," leading examples of artificial intelligence in dentistry.

This program provides information on 97 conditions and is designed to assist in the differential diagnosis of oral radiographs. The first publications using ANN in dentistry were done in PubMed in 1994. Over the last decade, however, there has been a significant increase in the number of applications of machine learning and deep learning in dentistry.

Results

Deep learning is helpful in the following ways:

-orthodontics, for the automated detection of landmarks and for the prediction of growth and development;

-cardiology, primarily for the analysis of radiographs for dental caries but also for caries risk assessment;

-periodontology and implantology, mainly for the analysis of radiographs but also for the evaluation of periodontal risk;

-endodontics, for the detection of endodontic pathologies from radiographs and for the prediction of the outcome of root canal treatment;

-oral and maxillofacial surgery to assist in surgical planning;

-in general, in dento-maxillofacial radiology, for image analysis (for example, to detect osteonecrosis of the jaw, temporomandibular disorders, or oral squamous cell carcinoma).

Conclusions

From the data of this review, which must be confirmed in other similar studies and reviews, it can be concluded that researchers and physicians can benefit from this study by gaining insights into artificial intelligence and deep learning, computer systems that will characterize the future doctor and dentist.