Potential of GPT-4 for Detecting Errors in Radiology Reports : Implications for Reporting Accuracy
Background Errors in radiology reports may occur because of resident-to-attending discrepancies, speech recognition inaccuracies, and large workload. Large language models, such as GPT-4 (ChatGPT; OpenAI), may assist in generating reports. Purpose To assess effectiveness of GPT-4 in identifying common errors in radiology reports, focusing on performance, time, and cost-efficiency. Materials and Methods In this retrospective study, 200 radiology reports (radiography and cross-sectional imaging [CT and MRI]) were compiled between June 2023 and December 2023 at one institution. There were 150 errors from five common error categories (omission, insertion, spelling, side confusion, and other) intentionally inserted into 100 of the reports and used as the reference standard. Six radiologists (two senior radiologists, two attending physicians, and two residents) and GPT-4 were tasked with detecting these errors. Overall error detection performance, error detection in the five error categories, and reading time were assessed using Wald χ2 tests and paired-sample t tests. Results GPT-4 (detection rate, 82.7%;124 of 150; 95% CI: 75.8, 87.9) matched the average detection performance of radiologists independent of their experience (senior radiologists, 89.3% [134 of 150; 95% CI: 83.4, 93.3]; attending physicians, 80.0% [120 of 150; 95% CI: 72.9, 85.6]; residents, 80.0% [120 of 150; 95% CI: 72.9, 85.6]; P value range, .522-.99). One senior radiologist outperformed GPT-4 (detection rate, 94.7%; 142 of 150; 95% CI: 89.8, 97.3; P = .006). GPT-4 required less processing time per radiology report than the fastest human reader in the study (mean reading time, 3.5 seconds ± 0.5 [SD] vs 25.1 seconds ± 20.1, respectively; P < .001; Cohen d = -1.08). The use of GPT-4 resulted in lower mean correction cost per report than the most cost-efficient radiologist ($0.03 ± 0.01 vs $0.42 ± 0.41; P < .001; Cohen d = -1.12). Conclusion The radiology report error detection rate of GPT-4 was comparable with that of radiologists, potentially reducing work hours and cost. © RSNA, 2024 See also the editorial by Forman in this issue.
| Medienart: |
E-Artikel |
|---|
| Erscheinungsjahr: |
2024 |
|---|---|
| Erschienen: |
2024 |
| Enthalten in: |
Zur Gesamtaufnahme - volume:311 |
|---|---|
| Enthalten in: |
Radiology - 311(2024), 1 vom: 01. Apr., Seite e232714 |
| Sprache: |
Englisch |
|---|
| Beteiligte Personen: |
Gertz, Roman Johannes [VerfasserIn] |
|---|
| Links: |
|---|
| Themen: |
|---|
| Anmerkungen: |
Date Completed 17.04.2024 Date Revised 17.04.2024 published: Print Citation Status MEDLINE |
|---|
| doi: |
10.1148/radiol.232714 |
|---|
| funding: |
|
|---|---|
| Förderinstitution / Projekttitel: |
|
| PPN (Katalog-ID): |
NLM371141249 |
|---|
| LEADER | 01000caa a22002652 4500 | ||
|---|---|---|---|
| 001 | NLM371141249 | ||
| 003 | DE-627 | ||
| 005 | 20240417233133.0 | ||
| 007 | cr uuu---uuuuu | ||
| 008 | 240416s2024 xx |||||o 00| ||eng c | ||
| 024 | 7 | |a 10.1148/radiol.232714 |2 doi | |
| 028 | 5 | 2 | |a pubmed24n1378.xml |
| 035 | |a (DE-627)NLM371141249 | ||
| 035 | |a (NLM)38625012 | ||
| 040 | |a DE-627 |b ger |c DE-627 |e rakwb | ||
| 041 | |a eng | ||
| 100 | 1 | |a Gertz, Roman Johannes |e verfasserin |4 aut | |
| 245 | 1 | 0 | |a Potential of GPT-4 for Detecting Errors in Radiology Reports |b Implications for Reporting Accuracy |
| 264 | 1 | |c 2024 | |
| 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 17.04.2024 | ||
| 500 | |a Date Revised 17.04.2024 | ||
| 500 | |a published: Print | ||
| 500 | |a Citation Status MEDLINE | ||
| 520 | |a Background Errors in radiology reports may occur because of resident-to-attending discrepancies, speech recognition inaccuracies, and large workload. Large language models, such as GPT-4 (ChatGPT; OpenAI), may assist in generating reports. Purpose To assess effectiveness of GPT-4 in identifying common errors in radiology reports, focusing on performance, time, and cost-efficiency. Materials and Methods In this retrospective study, 200 radiology reports (radiography and cross-sectional imaging [CT and MRI]) were compiled between June 2023 and December 2023 at one institution. There were 150 errors from five common error categories (omission, insertion, spelling, side confusion, and other) intentionally inserted into 100 of the reports and used as the reference standard. Six radiologists (two senior radiologists, two attending physicians, and two residents) and GPT-4 were tasked with detecting these errors. Overall error detection performance, error detection in the five error categories, and reading time were assessed using Wald χ2 tests and paired-sample t tests. Results GPT-4 (detection rate, 82.7%;124 of 150; 95% CI: 75.8, 87.9) matched the average detection performance of radiologists independent of their experience (senior radiologists, 89.3% [134 of 150; 95% CI: 83.4, 93.3]; attending physicians, 80.0% [120 of 150; 95% CI: 72.9, 85.6]; residents, 80.0% [120 of 150; 95% CI: 72.9, 85.6]; P value range, .522-.99). One senior radiologist outperformed GPT-4 (detection rate, 94.7%; 142 of 150; 95% CI: 89.8, 97.3; P = .006). GPT-4 required less processing time per radiology report than the fastest human reader in the study (mean reading time, 3.5 seconds ± 0.5 [SD] vs 25.1 seconds ± 20.1, respectively; P < .001; Cohen d = -1.08). The use of GPT-4 resulted in lower mean correction cost per report than the most cost-efficient radiologist ($0.03 ± 0.01 vs $0.42 ± 0.41; P < .001; Cohen d = -1.12). Conclusion The radiology report error detection rate of GPT-4 was comparable with that of radiologists, potentially reducing work hours and cost. © RSNA, 2024 See also the editorial by Forman in this issue | ||
| 650 | 4 | |a Journal Article | |
| 700 | 1 | |a Dratsch, Thomas |e verfasserin |4 aut | |
| 700 | 1 | |a Bunck, Alexander Christian |e verfasserin |4 aut | |
| 700 | 1 | |a Lennartz, Simon |e verfasserin |4 aut | |
| 700 | 1 | |a Iuga, Andra-Iza |e verfasserin |4 aut | |
| 700 | 1 | |a Hellmich, Martin Gunnar |e verfasserin |4 aut | |
| 700 | 1 | |a Persigehl, Thorsten |e verfasserin |4 aut | |
| 700 | 1 | |a Pennig, Lenhard |e verfasserin |4 aut | |
| 700 | 1 | |a Gietzen, Carsten Herbert |e verfasserin |4 aut | |
| 700 | 1 | |a Fervers, Philipp |e verfasserin |4 aut | |
| 700 | 1 | |a Maintz, David |e verfasserin |4 aut | |
| 700 | 1 | |a Hahnfeldt, Robert |e verfasserin |4 aut | |
| 700 | 1 | |a Kottlors, Jonathan |e verfasserin |4 aut | |
| 773 | 0 | 8 | |i Enthalten in |t Radiology |d 1945 |g 311(2024), 1 vom: 01. Apr., Seite e232714 |w (DE-627)NLM000054860 |x 1527-1315 |7 nnns |
| 773 | 1 | 8 | |g volume:311 |g year:2024 |g number:1 |g day:01 |g month:04 |g pages:e232714 |
| 856 | 4 | 0 | |u http://dx.doi.org/10.1148/radiol.232714 |3 Volltext |
| 912 | |a GBV_USEFLAG_A | ||
| 912 | |a GBV_NLM | ||
| 951 | |a AR | ||
| 952 | |d 311 |j 2024 |e 1 |b 01 |c 04 |h e232714 | ||