Artificial intelligence, machine learning, and deep learning in rhinology: a systematic review

Abstract

Purpose This PRISMA-compliant systematic review aims to analyze the existing applications of artificial intelligence (AI), machine learning, and deep learning for rhinological purposes and compare works in terms of data pool size, AI systems, input and outputs, and model reliability. Methods MEDLINE, Embase, Web of Science, Cochrane Library, and ClinicalTrials.gov databases. Search criteria were designed to include all studies published until December 2021 presenting or employing AI for rhinological applications. We selected all original studies specifying AI models reliability. After duplicate removal, abstract and full-text selection, and quality assessment, we reviewed eligible articles for data pool size, AI tools used, input and outputs, and model reliability. Results Among 1378 unique citations, 39 studies were deemed eligible. Most studies (n = 29) were technical papers. Input included compiled data, verbal data, and 2D images, while outputs were in most cases dichotomous or selected among nominal classes. The most frequently employed AI tools were support vector machine for compiled data and convolutional neural network for 2D images. Model reliability was variable, but in most cases was reported to be between 80% and 100%. Conclusions AI has vast potential in rhinology, but an inherent lack of accessible code sources does not allow for sharing results and advancing research without reconstructing models from scratch. While data pools do not necessarily represent a problem for model construction, presently available tools appear limited in allowing employment of raw clinical data, thus demanding immense interpretive work prior to the analytic process.

Applying Artificial Intelligence to Identify Common Targets for Treatment of Asthma, Eczema, and Food Allergy

“…AI is revolutionizing the pharmaceutical industry by accelerating every single step of drug development, from the discovery of potential therapeutic targets (8), prediction of protein structures (43), generation of hit compounds (44), to the prediction of drug safety (45) and efficacy (46). The uses of AI in allergy prediction, diagnosis, and medicine were also documented (47,48,49). For instance, esophageal mRNA transcripts were analyzed by machine learning strategies (i.e., weighted factor analysis followed by random forest classification) to predict the probability of developing eosinophilic esophagitis during food allergy response with high sensitivity and specificity (50).…”

International Consensus Statement on Allergy and Rhinology: Sinonasal Tumors

“…Traditional staging systems also may not capture all meaningful prognostic information, and there is a need to rethink and potentially redesign informative staging systems that integrate these critical features (e.g., Hyams grade in ONB, mutational burden). With the rise of artificial intelligence applications in medicine as a whole, systematic assessment of unique indicators of tumor behavior based on radiology and pathology may emerge 2220–2222 . We identify as ongoing research needs in a tumor‐specific manner: Identification of new/alternative strategies, biomarkers, and imaging modalities to more accurately diagnose patients with sinonasal tumors, especially malignancies at an earlier stage. Development of enhanced imaging modalities that evaluate the extent of involvement of sinonasal tumors, in particular orbital and intracranial involvement.…”

Are artificial intelligence large language models a reliable tool for difficult differential diagnosis? An a posteriori analysis of a peculiar case of necrotizing otitis externa

“…4 To date, the majority of AI studies in medicine focus on its applications in radiology, including in the head and neck region. 5,6 Chat Generative Pre-Trained Transformer (ChatGPT), an example of a large language model (LLM), is an easily accessible and user-friendly AI tool that has gained recent attention due to its ability to textually interact with near-human capability, due to the tool's extensive training on a wide range of texts available on the internet. Despite its apparent ease of use and unlimited capabilities, this tool application has been only minimally explored, particularly in more niche settings such as otolaryngology.…”

“…Potential medical applications of such tools, to assist either in diagnosis, therapeutic decisions, and prediction of outcomes, have attracted considerable interest since its early foundations in the middle of the last century 4 . To date, the majority of AI studies in medicine focus on its applications in radiology, including in the head and neck region 5,6 …”