Background Sport related concussion (SRC) is an acute and transient clinical condition that affects athletes who have suffered a mild cranio-encephalic trauma. The clinical management of a patient diagnosed with SRC must be targeted and personalized to allow complete recovery and safety and at the same time as rapid return to the field as possible. Aims To evaluate the current state of the art of SRC in association with the use of AI, focusing on some domains such as diagnosis, imaging, monitoring strategies and prognosis. This goal can be achieved by comparing data, such as sensitivity and specificity, of traditional methods against methods using AI. The aim of the study is to determine whether AI can guarantee better results than traditional methods and especially in which domains it is more promising, thus contributing to a more accurate, personalized and efficient management of SRC. Methods A scoping review was conducted, using a string to find relevant articles from six separate databases: PubMed, EMBASE, SPORTDiscus, Scopus, Web of Science, and Cochrane. After finalizing and adapting the search for each database, duplicates were removed. Two authors independently selected relevant studies for the scoping review, and in case of disagreement, a third author resolved the discrepancies. Results The database search generated 6577 results. Then, the screening and selection process of the articles was completed using Rayyan. A total of 38 articles were included in the scoping review. After an extensive examination of reference lists, two more papers were added, bringing the total number of articles included in the scoping review to 40. Finally, a summary table was prepared to summarize the main characteristics of each included study. For each domain, the main information was described, including: number of studies, country where the study was conducted, type of study, mean age of participants, male/female split, diagnostic tests used, type of AI used and comparison between traditional methods and AI methods. The results of AI were promising, reaching high accuracy in the detection of SRC. The diagnostic techniques to which AI was applied are numerous, such as EEG, voice analysis, use of wearable sensors and MRI. Not all the studies included in the scoping review presented a comparison with traditional methods, but when the comparison was present, AI was almost always able to obtain better results. Conclusions These results suggest that integrating AI into the management of SRC can improve diagnostic accuracy, provide more timely monitoring, and personalize treatment for each patient. Further large-scale studies are needed to promote their application in clinical practice. In conclusion, AI therefore represents a concrete opportunity that should not replace traditional methods but rather support them. Furthermore, its use must be guided by robust scientific evidence.
Presupposti dello studio La sport related concussion (SRC) è una condizione clinica acuta e transitoria che colpisce gli atleti che hanno subito un trauma cranio-encefalico lieve. La gestione clinica di un paziente con diagnosi di SRC deve essere mirata e personalizzata per permettere una guarigione completa ed un ritorno in campo sicuro e al tempo stesso più rapido possibile. Scopo dello studio Valutare lo stato dell’arte attuale della SRC in associazione all’utilizzo dell’AI, focalizzandosi su alcuni domini come diagnosi, imaging, strategie di monitoraggio e prognosi. Questo obiettivo è realizzabile comparando dati, come la sensibilità e la specificità, di metodi tradizionali contro metodi che utilizzano AI. Lo scopo dello studio è determinare se l’AI può garantire risultati migliori dei metodi tradizionali e soprattutto in quali domini risulti più promettente, contribuendo così a una gestione più accurata, personalizzata ed efficiente della SRC. Metodi È stata condotta una scoping review, utilizzando una stringa per trovare articoli rilevanti da sei distinti database: PubMed, EMBASE, SPORTDiscus, Scopus, Web of Science, and Cochrane. Dopo aver finalizzato e adattato la ricerca per ogni database, i duplicati sono stati rimossi. Due autori hanno selezionato in modo indipendente gli studi rilevanti per la scoping review e, in caso di divergenza, un terzo autore ha risolto le discrepanze. Risultati La ricerca su database ha generato 6577 risultati. In seguito, il processo di screening e selezione degli articoli è stato ultimato usando Rayyan. Un totale di 38 articoli è stato incluso nella scoping review. Dopo un esame approfondito degli elenchi dei riferimenti bibliografici, sono stati aggiunti altri due articoli, portando il numero totale degli articoli inclusi nella scoping review a 40. Infine, è stata elaborata una tabella sintetica per riassumere le principali caratteristiche di ciascuno studio incluso. Per ogni dominio sono state descritte le informazioni principali tra cui: numero degli studi, nazione in cui si è svolto lo studio, tipologia dello studio, età media dei partecipanti, divisione uomini/donne, test diagnostici utilizzati, tipo di AI usata e confronto tra metodi tradizionali e metodi AI. I risultati dell’AI sono stati promettenti raggiungendo un’elevata accuratezza nella rilevazione di SRC. Le tecniche diagnostiche a cui è stata applicata l’AI sono numerose, ad esempio EEG, analisi della voce, uso di sensori indossabili ed RMI. Non tutti gli studi compresi nella scoping review presentavano un confronto con metodi tradizionali, ma quando il confronto era presente quasi sempre l’AI è stata in grado di ottenere dei risultati migliori. Conclusioni Questi risultati suggeriscono che l’integrazione dell’AI nella gestione della SRC può migliorare l’accuratezza diagnostica, fornire un monitoraggio più tempestivo e personalizzare il trattamento per ogni paziente. Per favorire l’applicazione nella pratica clinica sono necessari altri studi su larga scala. Per concludere, l’AI rappresenta quindi un’opportunità concreta che non deve sostituire i metodi tradizionali, ma piuttosto affiancarli. Inoltre, il suo utilizzo deve essere guidato da evidenze scientifiche robuste.
Artificial Intelligence application in sport concussion: current development and future directions
DAL PONTE, FILIPPO
2024/2025
Abstract
Background Sport related concussion (SRC) is an acute and transient clinical condition that affects athletes who have suffered a mild cranio-encephalic trauma. The clinical management of a patient diagnosed with SRC must be targeted and personalized to allow complete recovery and safety and at the same time as rapid return to the field as possible. Aims To evaluate the current state of the art of SRC in association with the use of AI, focusing on some domains such as diagnosis, imaging, monitoring strategies and prognosis. This goal can be achieved by comparing data, such as sensitivity and specificity, of traditional methods against methods using AI. The aim of the study is to determine whether AI can guarantee better results than traditional methods and especially in which domains it is more promising, thus contributing to a more accurate, personalized and efficient management of SRC. Methods A scoping review was conducted, using a string to find relevant articles from six separate databases: PubMed, EMBASE, SPORTDiscus, Scopus, Web of Science, and Cochrane. After finalizing and adapting the search for each database, duplicates were removed. Two authors independently selected relevant studies for the scoping review, and in case of disagreement, a third author resolved the discrepancies. Results The database search generated 6577 results. Then, the screening and selection process of the articles was completed using Rayyan. A total of 38 articles were included in the scoping review. After an extensive examination of reference lists, two more papers were added, bringing the total number of articles included in the scoping review to 40. Finally, a summary table was prepared to summarize the main characteristics of each included study. For each domain, the main information was described, including: number of studies, country where the study was conducted, type of study, mean age of participants, male/female split, diagnostic tests used, type of AI used and comparison between traditional methods and AI methods. The results of AI were promising, reaching high accuracy in the detection of SRC. The diagnostic techniques to which AI was applied are numerous, such as EEG, voice analysis, use of wearable sensors and MRI. Not all the studies included in the scoping review presented a comparison with traditional methods, but when the comparison was present, AI was almost always able to obtain better results. Conclusions These results suggest that integrating AI into the management of SRC can improve diagnostic accuracy, provide more timely monitoring, and personalize treatment for each patient. Further large-scale studies are needed to promote their application in clinical practice. In conclusion, AI therefore represents a concrete opportunity that should not replace traditional methods but rather support them. Furthermore, its use must be guided by robust scientific evidence.| File | Dimensione | Formato | |
|---|---|---|---|
|
Dal_Ponte_Filippo.pdf
accesso aperto
Dimensione
2.6 MB
Formato
Adobe PDF
|
2.6 MB | Adobe PDF | Visualizza/Apri |
The text of this website © Università degli studi di Padova. Full Text are published under a non-exclusive license. Metadata are under a CC0 License
https://hdl.handle.net/20.500.12608/86495