Anh N. Le

Anh N. Le

Auto-generated profile, based on publicly available information

Anh N. Le

Division of Hematology-Oncology, Department of Medicine, Perelman School of Medicine, University of Pennsylvania, Philadelphia, Pennsylvania, USA

Research Focus & Metrics

Academic Career Length

6Years (2021-Present)

(Calculated once Scilit publication data is synced: current year − year of first publication + 1)

Representative Publications

View All Publications
2 July 2021 in JMIR Cancer
Verified
Journal Article
Open access
SYShun Yu
Shun Yu
ALAnh Le
Anh Le
EFEmily Feld
Emily Feld
ESEmily Schriver
Emily Schriver
PGPeter Gabriel
Peter Gabriel
DMDanielle Mowery
Danielle Mowery
Background: Natural language processing (NLP) offers significantly faster variable extraction compared to traditional human extraction but cannot interpret complicated notes as well as humans can. Thus, we hypothesized that an “NLP-assisted” extraction system, which uses humans for complicated notes and NLP for uncomplicated notes, could produce faster extraction without compromising accuracy. Objective: The aim of this study was to develop and pilot an NLP-assisted extraction system to leverage the strengths of both human and NLP extraction of prostate cancer Gleason scores. Methods: We collected all available clinical and pathology notes for prostate cancer patients in an unselected academic biobank cohort. We developed an NLP system to extract prostate cancer Gleason scores from both clinical and pathology notes. Next, we designed and implemented the NLP-assisted extraction system algorithm to categorize notes into “uncomplicated” and “complicated” notes. Uncomplicated notes were assigned to NLP extraction and complicated notes were assigned to human extraction. We randomly reviewed 200 patients to assess the accuracy and speed of our NLP-assisted extraction system and compared it to NLP extraction alone and human extraction alone. Results: Of the 2051 patients in our cohort, the NLP system extracted a prostate surgery Gleason score from 1147 (55.92%) patients and a prostate biopsy Gleason score from 1624 (79.18%) patients. Our NLP-assisted extraction system had an overall accuracy rate of 98.7%, which was similar to the accuracy of human extraction alone (97.5%; P=.17) and significantly higher than the accuracy of NLP extraction alone (95.3%; P<.001). Moreover, our NLP-assisted extraction system reduced the workload of human extractors by approximately 95%, resulting in an average extraction time of 12.7 seconds per patient (vs 256.1 seconds per patient for human extraction alone). Conclusions: We demonstrated that an NLP-assisted extraction system was able to achieve much faster Gleason score extraction compared to traditional human extraction without sacrificing accuracy.

7 July 2023 in Cancer Prevention Research
Verified
Journal Article
Closed access
PDF file
LKLaura A.T. Kagami
Laura A.T. Kagami
YDYun K. Du
Yun K. Du
CFConrad J. Fernandes
Conrad J. Fernandes
ALAnh N. Le
Anh N. Le
MGMadeline Good
Madeline Good
KMKara N Maxwell
Kara N Maxwell
Li-Fraumeni Syndrome (LFS) is a hereditary cancer predisposition syndrome with up to 90% lifetime cancer risk. Cancer screening, including annual whole-body MRI (WB-MRI), is recommended due to known survival advantage, with cancer detection rate of 7% on initial screening. Intervention and cancer detection rates on subsequent screenings are unknown. Clinical data for pediatric and adult patients with LFS (n = 182) were reviewed, including instances of WB-MRI screening and interventions based on screening results. For each WB-MRI screening, interventions including biopsy and secondary imaging, as well as rate of cancer diagnosis, were analyzed comparing initial versus subsequent WB-MRI. Of the total cohort (n = 182), we identified 68 adult patients and 50 pediatric patients who had undergone at least two WB-MRI screenings, with a mean of 3.8 ± 1.9 (adults) and 4.0 ± 2.1 (pediatric) screenings. Findings on initial screening led to an imaging or invasive intervention in 38% of adults and 20% of children. On follow up, overall intervention rates were lower for adults (19%, P = 0.0026) and stable for children (19%, P = NS). Thirteen cancers were detected overall (7% of adult and 14% of pediatric scans), on both initial (pediatric: 4%, adult: 3%) and subsequent (pediatric: 10%, adult: 6%) screenings. Rates of intervention after WB-MRI screening decreased significantly in adults between first and subsequent exams and remained stable in pediatric patients. Cancer detection rates were similar on screening (3%–4% initial, 6%–10% subsequent) for both children and adults. These findings provide important data for counseling patients with LFS about screening outcomes. Prevention Relevance: The cancer detection rate, burden of recommended interventions, and rate of false-positive findings found on subsequent WB-MRI screenings in patients with LFS are not well understood. Our findings suggest that annual WB-MRI screening has clinical utility and likely does not result in an unnecessary invasive intervention burden for patients.

14 July 2022 in Ecotoxicology
Verified
Journal Article
Closed access
PDF file
CMChristopher J. McLimans
Christopher J. McLimans
KSKatharine Shelledy
Katharine Shelledy
WCWilliam Conrad
William Conrad
KPKyle Prendergast
Kyle Prendergast
ALAnh N. Le
Anh N. Le
VBVincent P. Buonaccorsi
Vincent P. Buonaccorsi

21 November 2021 in medRxiv
Verified
Preprint
Open access
PDF file
ADAnahita Davoudi
Anahita Davoudi
SYShun Yu
Shun Yu
ADAbigail Doucette
Abigail Doucette
PGPeter Gabriel
Peter Gabriel
MMMark Miller
Mark Miller
DMDanielle L. Mowery
Danielle L. Mowery
Although NLP has been used to support cancer research more broadly, the development of NLP algorithms to extract evidence of progression from clinical notes to support lung cancer research is still in its infancy. In this study, we trained supervised machine learning classifiers using rich semantic features to detect and classify statements of progression status from radiology exams. Our progression status classifier achieves high F1-scores for detecting and discerning progression (0.80), stable (0.82), and not relevant (0.92) sentences, demonstrating promising performance. We are actively integrating these extractions with structured electronic health record data using ontologies to instantiate a longitudinal model of progression among non-small cell lung cancer patients.

25 February 2026 in American Journal of Gastroenterology
Verified
Journal Article
Closed access
SSSophia R. Spiegel
Sophia R. Spiegel
COCaitlin Orr
Caitlin Orr
NENicolas U. Edgar
Nicolas U. Edgar
SCSophie R. Cahill
Sophie R. Cahill
ALAnh N. Le
Anh N. Le
KMKara N Maxwell
Kara N Maxwell
INTRODUCTION: Individuals with Li-Fraumeni syndrome (LFS), a cancer predisposition syndrome caused by TP53 pathogenic germline variants (PGV), have an increased risk of colorectal cancer (CRC). However, limited data exist on colonoscopy metrics, including total neoplasia detection rates, in patients with LFS. METHODS: We conducted a retrospective cohort study of colorectal neoplasia incidence and characteristics in 663 individuals with LFS. We determined total neoplasia detection rate, adenoma detection rate (ADR), advanced ADR, serrated lesion detection rate (SDR), and CRC detection rate, stratified by age, sex, and PGV subtype, in 311 surveillance colonoscopies from 206 adults performed January 2019 to August 2024. RESULTS: CRC was reported in 4.5% of 663 adults and 0.8% of 124 pediatric patients with LFS. Among 206 adults undergoing colonoscopy, total neoplasia detection rate, ADR, SDR, advanced precancerous polyp detection rate, and CRC detection rate were 37%, 27%, 9.3%, 4.5%, and 0.64%, respectively. Detection rates were similar between loss of function and hypomorphic PGVs. Adults aged 45–75 had higher total neoplasia detection rate, ADR, advanced ADR, SDR, and CRC detection rate than those aged 25–45. Male sex was associated with increased total neoplasia detection rate (53.7% vs 30.7%, P = 0.0004), ADR (41.2% vs 22.1%, P = 0.0013), and advanced ADR (7.5% vs 1.3%, P = 0.0105). DISCUSSION: Patients with LFS have comparable ADR and higher SDR with average-risk adults despite earlier, more frequent surveillance, irrespective of TP53 PGV subtype. Our data support current recommendations for earlier and more frequent colonoscopy surveillance than average-risk guidelines, across LFS phenotypes.

15 February 2021 in JMIR Preprints
Verified
Preprint
Open access
SYShun Yu
Shun Yu
ALAnh Le
Anh Le
EFEmily Feld
Emily Feld
ESEmily Schriver
Emily Schriver
PGPeter Gabriel
Peter Gabriel
DMDanielle Mowery
Danielle Mowery
BACKGROUND Natural language processing (NLP) offers significantly faster variable extraction compared to traditional human extraction but cannot interpret complicated notes as well as humans can. Thus, we hypothesized that an “NLP-assisted” extraction system, which uses humans for complicated notes and NLP for uncomplicated notes, could produce faster extraction without compromising accuracy. OBJECTIVE The aim of this study was to develop and pilot an NLP-assisted extraction system to leverage the strengths of both human and NLP extraction of prostate cancer Gleason scores. METHODS We collected all available clinical and pathology notes for prostate cancer patients in an unselected academic biobank cohort. We developed an NLP system to extract prostate cancer Gleason scores from both clinical and pathology notes. Next, we designed and implemented the NLP-assisted extraction system algorithm to categorize notes into “uncomplicated” and “complicated” notes. Uncomplicated notes were assigned to NLP extraction and complicated notes were assigned to human extraction. We randomly reviewed 200 patients to assess the accuracy and speed of our NLP-assisted extraction system and compared it to NLP extraction alone and human extraction alone. RESULTS Of the 2051 patients in our cohort, the NLP system extracted a prostate surgery Gleason score from 1147 (55.92%) patients and a prostate biopsy Gleason score from 1624 (79.18%) patients. Our NLP-assisted extraction system had an overall accuracy rate of 98.7%, which was similar to the accuracy of human extraction alone (97.5%; P=.17) and significantly higher than the accuracy of NLP extraction alone (95.3%; P<.001). Moreover, our NLP-assisted extraction system reduced the workload of human extractors by approximately 95%, resulting in an average extraction time of 12.7 seconds per patient (vs 256.1 seconds per patient for human extraction alone). CONCLUSIONS We demonstrated that an NLP-assisted extraction system was able to achieve much faster Gleason score extraction compared to traditional human extraction without sacrificing accuracy.