Artificial intelligence in medicine : are we ready?
Artificial intelligence (Ai) tools, in particular those based on machine learning (Ml) and deep learning techniques, have found application in numerous fields of medicine. They are used to create chatbots to provide health information to citizens and patients, to make medical diagnoses, to identify the best treatment to use for a specific patient, to predict a patient's risk of experiencing a specific event, to identify the most promising drug candidate in which to invest in clinical research, and to identify possible cause-effect relationships between pathologies and data collected through the various available sources. The results of published clinical trials measuring the reliability, safety, and efficacy of these tools are often encouraging. However, there are systematic reviews and meta-analyses highlighting the methodological limitations of these studies. Many of these are retrospective and based on previously assembled datasets, while few are prospective ones conducted in real clinical settings and very few are those based on randomized controlled clinical trials. Furthermore, many of these adopt an internal validation of the Ai system to the detriment of an external validation, while the comparison between the performance of the Ml models and that of the experts is only in a few cases conducted using the same dataset. Before being used in the healthcare field, Ai and Ml systems should pass through the scrutiny of rigorous scientific validation, based on methodologically solid studies (prospective, possibly randomized and conducted in real clinical environments) which demonstrate non-inferiority, or superiority, as well as cost-effectiveness, compared to the conventional diagnostic and decision-making pathway. Furthermore, it is necessary to demonstrate the safety and reproducibility in the use of the software and to consider the emerging ethical and legal issues inherent in the professional liability of the doctor in the interaction with the algorithms.
Medienart: |
E-Artikel |
---|
Erscheinungsjahr: |
2023 |
---|---|
Erschienen: |
2023 |
Enthalten in: |
Zur Gesamtaufnahme - volume:114 |
---|---|
Enthalten in: |
Recenti progressi in medicina - 114(2023), 3 vom: 01. März, Seite 142-144 |
Sprache: |
Italienisch |
---|
Weiterer Titel: |
Intelligenza artificiale in medicina: siamo pronti? |
---|
Beteiligte Personen: |
Santoro, Eugenio [VerfasserIn] |
---|
Links: |
---|
Themen: |
---|
Anmerkungen: |
Date Completed 24.02.2023 Date Revised 03.03.2023 published: Print Citation Status MEDLINE |
---|
doi: |
10.1701/3981.39636 |
---|
funding: |
|
---|---|
Förderinstitution / Projekttitel: |
|
PPN (Katalog-ID): |
NLM353216771 |
---|
LEADER | 01000naa a22002652 4500 | ||
---|---|---|---|
001 | NLM353216771 | ||
003 | DE-627 | ||
005 | 20231226055509.0 | ||
007 | cr uuu---uuuuu | ||
008 | 231226s2023 xx |||||o 00| ||ita c | ||
024 | 7 | |a 10.1701/3981.39636 |2 doi | |
028 | 5 | 2 | |a pubmed24n1177.xml |
035 | |a (DE-627)NLM353216771 | ||
035 | |a (NLM)36815415 | ||
040 | |a DE-627 |b ger |c DE-627 |e rakwb | ||
041 | |a ita | ||
100 | 1 | |a Santoro, Eugenio |e verfasserin |4 aut | |
245 | 1 | 0 | |a Artificial intelligence in medicine |b are we ready? |
246 | 3 | 3 | |a Intelligenza artificiale in medicina: siamo pronti? |
264 | 1 | |c 2023 | |
336 | |a Text |b txt |2 rdacontent | ||
337 | |a ƒaComputermedien |b c |2 rdamedia | ||
338 | |a ƒa Online-Ressource |b cr |2 rdacarrier | ||
500 | |a Date Completed 24.02.2023 | ||
500 | |a Date Revised 03.03.2023 | ||
500 | |a published: Print | ||
500 | |a Citation Status MEDLINE | ||
520 | |a Artificial intelligence (Ai) tools, in particular those based on machine learning (Ml) and deep learning techniques, have found application in numerous fields of medicine. They are used to create chatbots to provide health information to citizens and patients, to make medical diagnoses, to identify the best treatment to use for a specific patient, to predict a patient's risk of experiencing a specific event, to identify the most promising drug candidate in which to invest in clinical research, and to identify possible cause-effect relationships between pathologies and data collected through the various available sources. The results of published clinical trials measuring the reliability, safety, and efficacy of these tools are often encouraging. However, there are systematic reviews and meta-analyses highlighting the methodological limitations of these studies. Many of these are retrospective and based on previously assembled datasets, while few are prospective ones conducted in real clinical settings and very few are those based on randomized controlled clinical trials. Furthermore, many of these adopt an internal validation of the Ai system to the detriment of an external validation, while the comparison between the performance of the Ml models and that of the experts is only in a few cases conducted using the same dataset. Before being used in the healthcare field, Ai and Ml systems should pass through the scrutiny of rigorous scientific validation, based on methodologically solid studies (prospective, possibly randomized and conducted in real clinical environments) which demonstrate non-inferiority, or superiority, as well as cost-effectiveness, compared to the conventional diagnostic and decision-making pathway. Furthermore, it is necessary to demonstrate the safety and reproducibility in the use of the software and to consider the emerging ethical and legal issues inherent in the professional liability of the doctor in the interaction with the algorithms | ||
650 | 4 | |a English Abstract | |
650 | 4 | |a Journal Article | |
773 | 0 | 8 | |i Enthalten in |t Recenti progressi in medicina |d 1948 |g 114(2023), 3 vom: 01. März, Seite 142-144 |w (DE-627)NLM000023124 |x 2038-1840 |7 nnns |
773 | 1 | 8 | |g volume:114 |g year:2023 |g number:3 |g day:01 |g month:03 |g pages:142-144 |
856 | 4 | 0 | |u http://dx.doi.org/10.1701/3981.39636 |3 Volltext |
912 | |a GBV_USEFLAG_A | ||
912 | |a GBV_NLM | ||
951 | |a AR | ||
952 | |d 114 |j 2023 |e 3 |b 01 |c 03 |h 142-144 |