Closing the gap in surveillance and audit of invasive mold diseases for antifungal stewardship using machine learning

Diva Baggio, Trisha Peel, Anton Y. Peleg, Sharon Avery, Madhurima Prayaga, Michelle Foo, Reza Haffari, Ming Liu, Christoph Bergmeir, Michelle Ananda-Rajah

Research output: Contribution to journalArticleResearchpeer-review

Abstract

Clinical audit of invasive mold disease (IMD) in hematology patients is inefficient due to the difficulties of case finding. This results in antifungal stewardship (AFS) programs preferentially reporting drug cost and consumption rather than measures that actually reflect quality of care. We used machine learning-based natural language processing (NLP) to non-selectively screen chest tomography (CT) reports for pulmonary IMD, verified by clinical review against international definitions and benchmarked against key AFS measures. NLP screened 3014 reports from 1 September 2008 to 31 December 2017, generating 784 positives that after review, identified 205 IMD episodes (44% probable-proven) in 185 patients from 50,303 admissions. Breakthrough-probable/proven-IMD on antifungal prophylaxis accounted for 60% of episodes with serum monitoring of voriconazole or posaconazole in the 2 weeks prior performed in only 53% and 69% of episodes, respectively. Fiberoptic bronchoscopy within 2 days of CT scan occurred in only 54% of episodes. The average turnaround of send-away bronchoalveolar galactomannan of 12 days (range 7–22) was associated with high empiric liposomal amphotericin consumption. A random audit of 10% negative reports revealed two clinically significant misses (0.9%, 2/223). This is the first successful use of applied machine learning for institutional IMD surveillance across an entire hematology population describing process and outcome measures relevant to AFS. Compared to current methods of clinical audit, semi-automated surveillance using NLP is more efficient and inclusive by avoiding restrictions based on any underlying hematologic condition, and has the added advantage of being potentially scalable
Original languageEnglish
Article number1390
Number of pages13
JournalJournal of Clinical Medicine
Volume8
Issue number9
DOIs
Publication statusPublished - 5 Sep 2019

Cite this

@article{ab5caa4a5a8346a6935af213a233316c,
title = "Closing the gap in surveillance and audit of invasive mold diseases for antifungal stewardship using machine learning",
abstract = "Clinical audit of invasive mold disease (IMD) in hematology patients is inefficient due to the difficulties of case finding. This results in antifungal stewardship (AFS) programs preferentially reporting drug cost and consumption rather than measures that actually reflect quality of care. We used machine learning-based natural language processing (NLP) to non-selectively screen chest tomography (CT) reports for pulmonary IMD, verified by clinical review against international definitions and benchmarked against key AFS measures. NLP screened 3014 reports from 1 September 2008 to 31 December 2017, generating 784 positives that after review, identified 205 IMD episodes (44{\%} probable-proven) in 185 patients from 50,303 admissions. Breakthrough-probable/proven-IMD on antifungal prophylaxis accounted for 60{\%} of episodes with serum monitoring of voriconazole or posaconazole in the 2 weeks prior performed in only 53{\%} and 69{\%} of episodes, respectively. Fiberoptic bronchoscopy within 2 days of CT scan occurred in only 54{\%} of episodes. The average turnaround of send-away bronchoalveolar galactomannan of 12 days (range 7–22) was associated with high empiric liposomal amphotericin consumption. A random audit of 10{\%} negative reports revealed two clinically significant misses (0.9{\%}, 2/223). This is the first successful use of applied machine learning for institutional IMD surveillance across an entire hematology population describing process and outcome measures relevant to AFS. Compared to current methods of clinical audit, semi-automated surveillance using NLP is more efficient and inclusive by avoiding restrictions based on any underlying hematologic condition, and has the added advantage of being potentially scalable",
author = "Diva Baggio and Trisha Peel and Peleg, {Anton Y.} and Sharon Avery and Madhurima Prayaga and Michelle Foo and Reza Haffari and Ming Liu and Christoph Bergmeir and Michelle Ananda-Rajah",
year = "2019",
month = "9",
day = "5",
doi = "10.3390/jcm8091390",
language = "English",
volume = "8",
journal = "Journal of Clinical Medicine",
issn = "2077-0383",
publisher = "MDPI",
number = "9",

}

Closing the gap in surveillance and audit of invasive mold diseases for antifungal stewardship using machine learning. / Baggio, Diva; Peel, Trisha; Peleg, Anton Y.; Avery, Sharon; Prayaga, Madhurima; Foo, Michelle; Haffari, Reza; Liu, Ming; Bergmeir, Christoph; Ananda-Rajah, Michelle.

In: Journal of Clinical Medicine, Vol. 8, No. 9, 1390, 05.09.2019.

Research output: Contribution to journalArticleResearchpeer-review

TY - JOUR

T1 - Closing the gap in surveillance and audit of invasive mold diseases for antifungal stewardship using machine learning

AU - Baggio, Diva

AU - Peel, Trisha

AU - Peleg, Anton Y.

AU - Avery, Sharon

AU - Prayaga, Madhurima

AU - Foo, Michelle

AU - Haffari, Reza

AU - Liu, Ming

AU - Bergmeir, Christoph

AU - Ananda-Rajah, Michelle

PY - 2019/9/5

Y1 - 2019/9/5

N2 - Clinical audit of invasive mold disease (IMD) in hematology patients is inefficient due to the difficulties of case finding. This results in antifungal stewardship (AFS) programs preferentially reporting drug cost and consumption rather than measures that actually reflect quality of care. We used machine learning-based natural language processing (NLP) to non-selectively screen chest tomography (CT) reports for pulmonary IMD, verified by clinical review against international definitions and benchmarked against key AFS measures. NLP screened 3014 reports from 1 September 2008 to 31 December 2017, generating 784 positives that after review, identified 205 IMD episodes (44% probable-proven) in 185 patients from 50,303 admissions. Breakthrough-probable/proven-IMD on antifungal prophylaxis accounted for 60% of episodes with serum monitoring of voriconazole or posaconazole in the 2 weeks prior performed in only 53% and 69% of episodes, respectively. Fiberoptic bronchoscopy within 2 days of CT scan occurred in only 54% of episodes. The average turnaround of send-away bronchoalveolar galactomannan of 12 days (range 7–22) was associated with high empiric liposomal amphotericin consumption. A random audit of 10% negative reports revealed two clinically significant misses (0.9%, 2/223). This is the first successful use of applied machine learning for institutional IMD surveillance across an entire hematology population describing process and outcome measures relevant to AFS. Compared to current methods of clinical audit, semi-automated surveillance using NLP is more efficient and inclusive by avoiding restrictions based on any underlying hematologic condition, and has the added advantage of being potentially scalable

AB - Clinical audit of invasive mold disease (IMD) in hematology patients is inefficient due to the difficulties of case finding. This results in antifungal stewardship (AFS) programs preferentially reporting drug cost and consumption rather than measures that actually reflect quality of care. We used machine learning-based natural language processing (NLP) to non-selectively screen chest tomography (CT) reports for pulmonary IMD, verified by clinical review against international definitions and benchmarked against key AFS measures. NLP screened 3014 reports from 1 September 2008 to 31 December 2017, generating 784 positives that after review, identified 205 IMD episodes (44% probable-proven) in 185 patients from 50,303 admissions. Breakthrough-probable/proven-IMD on antifungal prophylaxis accounted for 60% of episodes with serum monitoring of voriconazole or posaconazole in the 2 weeks prior performed in only 53% and 69% of episodes, respectively. Fiberoptic bronchoscopy within 2 days of CT scan occurred in only 54% of episodes. The average turnaround of send-away bronchoalveolar galactomannan of 12 days (range 7–22) was associated with high empiric liposomal amphotericin consumption. A random audit of 10% negative reports revealed two clinically significant misses (0.9%, 2/223). This is the first successful use of applied machine learning for institutional IMD surveillance across an entire hematology population describing process and outcome measures relevant to AFS. Compared to current methods of clinical audit, semi-automated surveillance using NLP is more efficient and inclusive by avoiding restrictions based on any underlying hematologic condition, and has the added advantage of being potentially scalable

U2 - 10.3390/jcm8091390

DO - 10.3390/jcm8091390

M3 - Article

VL - 8

JO - Journal of Clinical Medicine

JF - Journal of Clinical Medicine

SN - 2077-0383

IS - 9

M1 - 1390

ER -