Mākslīgā intelekta pielietojums lūzumu diagnostikā konvencionālā radiogrāfijā
No Thumbnail Available
Date
2023
Authors
Journal Title
Journal ISSN
Volume Title
Publisher
Rīgas Stradiņa universitāte
Rīga Stradiņš University
Rīga Stradiņš University
Abstract
Traumas rezultātā radušies kaulu lūzumi ir izplatītākais iemesls pacientu nonākšanai slimnīcu Neatliekamās medicīniskās palīdzības jeb Uzņemšanas nodaļā. Rutīnas izmeklējums kaulu lūzumu diagnostikā ir rentgenogrāfija. Ņemot vērā lielo radiologu, traumatologu, ortopēdu noslogojumu, pieaug risks uz kļūdām rentgenogrammu analizēšanā un gala diagnozes noteikšanā. Nosakot pacientiem nekorektu gala diagnozi, var tikt izvēlēta nepiemērota ārstēšana, kas var radīt nopietnas komplikācijas pacientu veselībai. Šobrīd ar vien plašāk medicīnas nozarē tiek ieviestas mākslīgā intelekta programmatūras, kas spēj diagnosticēt patoloģiskus veidojumus, kaulu lūzumus.
Šis ir retrospektīvs pētījums, kurā tika pētīts mākslīgā intelekta programmas AZmed “Rayvolve” specifiskums un jutība, kā arī Rīgas 2. slimnīcas un Traumatoloģijas un ortopēdijas slimnīcas traumatologu, ortopēdu, radiologu, radioloģijas un traumatoloģijas, ortopēdijas rezidentu viedoklis par mākslīgā intelekta programmu noderīgumu un precizitāti kaulu lūzumu diagnostikā.
Pētījuma rezultātā tika noteikts, ka mākslīgā intelekta programmas AZmed “Rayvolve” specifiskums ir 97,8% ,jutība 93,4% un negatīvā paredzamā vērtība ir 95,1%. Pētījumā tika secināt, ka programma ir pietiekoši precīza, lai varētu tikt izmantota kā asistents speciālistiem ikdienas kaulu lūzumu diagnostikā.
Anketēšanas rezultātos atspoguļojās speciālistu viedoklis par mākslīgā intelekta programmu precizitāti kaulu lūzumu diagnostikā, kur vislabāk tās tiek vērtētas garo kaulu lūzumu diagnostikā. Tika iegūts anketas dalībnieku viedoklis par programmu nepieciešamību visās Latvijas slimnīcu radioloģijas un Uzņemšanas nodaļās. Vairākums anketas dalībnieku jeb 72,2% atzīmēja, ka tas būtu nepieciešams. Taču speciālisti atzīmēja, ka neskatoties uz programmu noderīgumu, tām varētu būt negatīva ietekme uz speciālistu kvalifikāciju kaulu lūzumu diagnostikas jomā.
Mākslīgā intelekta programmas ir attīstības stadijā un tiek konstanti uzlabotas, lai spētu pēc iespējas precīzāk analizēt radioloģiskos izmeklējumus. Medicīnas nozarē, strauji notiekot digitalizācijai un mākslīgā intelekta programmu ieviešanai, svarīgi ir neaizmirst par pacientu personas datu aizsardzību, konfidencialitāti un kiberdrošību.
Tuvākajā nākotnē mākslīgā intelekta programmatūras vēl nebūs spējīgas aizstāt speciālistus, bet tās varētu kļūt par nozīmīgu patoloģiju, kaulu lūzumu diagnostikas daļu. Šādu programmu izmantošana varētu mazināt gala diagnozes noteikšanai patērēto laiku, mazināt pieļautās kļūdas radioloģisko izmeklējumu interpretācijā un samazināt speciālistu noslogojumu.
Traumatic bone fractures are the most common reason for patients to be admitted to hospital admissions. Radiography is a routine diagnostic test for bone fractures. Given the heavy workload of radiologists, traumatologists and orthopaedists, the risk of errors in analysing radiographs and making a final diagnosis is increasing. If the final diagnosis is incorrect, inappropriate treatment may be chosen, which may lead to serious complications for the patients' health. The medical sector is now increasingly adopting artificial intelligence software capable of diagnosing abnormal growths and bone fractures. This is a retrospective study that investigated the specificity and sensitivity of the artificial intelligence software AZmed "Rayvolve", as well as the opinions of traumatologists, orthopaedists, radiologists, radiology and traumatology, orthopaedics residents at Riga 2 Hospital and the Hospital of Traumatology and Orthopaedics on the usefulness and accuracy of artificial intelligence software in diagnosing bone fractures. The results determined that the specificity of the artificial intelligence programme AZmed "Rayvolve" was 97.8%, the sensitivity 93.4% and the negative predictive value 95.1%. The study concludes that the program is accurate enough to be used as an assistant for specialists in the routine diagnosis of bone fractures. The results of the questionnaire reflected the opinion of the experts on the accuracy of the artificial intelligence programmes in the diagnosis of bone fractures, where they are rated best in the diagnosis of long bone fractures. The questionnaire obtained the opinion of participants as to the necessity of the programs in all radiology and admission departments of Latvian hospitals. The majority of questionnaire participants, 72.2%, indicated that it would be necessary. However, specialists noted that despite the usefulness of the programmes, they could have a negative impact on the qualification of specialists in the field of bone fracture diagnosis. Artificial intelligence programmes are, however, under development and are constantly being improved to be able to analyse radiological examinations as accurately as possible. With the rapid digitisation of the medical sector and the introduction of artificial intelligence applications, it is important not to forget about the protection of patients' personal data, confidentiality and cybersecurity. Soon, artificial intelligence software will not yet be able to replace specialists, but it could become an important part of the diagnosis of pathologies and bone fractures. The use of such software could reduce the time taken to make a final diagnosis, reduce errors in the interpretation of radiological examinations and reduce the workload of specialists.
Traumatic bone fractures are the most common reason for patients to be admitted to hospital admissions. Radiography is a routine diagnostic test for bone fractures. Given the heavy workload of radiologists, traumatologists and orthopaedists, the risk of errors in analysing radiographs and making a final diagnosis is increasing. If the final diagnosis is incorrect, inappropriate treatment may be chosen, which may lead to serious complications for the patients' health. The medical sector is now increasingly adopting artificial intelligence software capable of diagnosing abnormal growths and bone fractures. This is a retrospective study that investigated the specificity and sensitivity of the artificial intelligence software AZmed "Rayvolve", as well as the opinions of traumatologists, orthopaedists, radiologists, radiology and traumatology, orthopaedics residents at Riga 2 Hospital and the Hospital of Traumatology and Orthopaedics on the usefulness and accuracy of artificial intelligence software in diagnosing bone fractures. The results determined that the specificity of the artificial intelligence programme AZmed "Rayvolve" was 97.8%, the sensitivity 93.4% and the negative predictive value 95.1%. The study concludes that the program is accurate enough to be used as an assistant for specialists in the routine diagnosis of bone fractures. The results of the questionnaire reflected the opinion of the experts on the accuracy of the artificial intelligence programmes in the diagnosis of bone fractures, where they are rated best in the diagnosis of long bone fractures. The questionnaire obtained the opinion of participants as to the necessity of the programs in all radiology and admission departments of Latvian hospitals. The majority of questionnaire participants, 72.2%, indicated that it would be necessary. However, specialists noted that despite the usefulness of the programmes, they could have a negative impact on the qualification of specialists in the field of bone fracture diagnosis. Artificial intelligence programmes are, however, under development and are constantly being improved to be able to analyse radiological examinations as accurately as possible. With the rapid digitisation of the medical sector and the introduction of artificial intelligence applications, it is important not to forget about the protection of patients' personal data, confidentiality and cybersecurity. Soon, artificial intelligence software will not yet be able to replace specialists, but it could become an important part of the diagnosis of pathologies and bone fractures. The use of such software could reduce the time taken to make a final diagnosis, reduce errors in the interpretation of radiological examinations and reduce the workload of specialists.
Description
Medicīna
Medicine
Veselības aprūpe
Health Care
Medicine
Veselības aprūpe
Health Care
Keywords
Radioloģija, Mākslīgais intelekts, kaulu lūzumi, Radiology, Artificial intelligence, bone fracture