Matthew S Alkaitis1,2, Monica N Agrawal1, Gregory J Riely3,4, Pedram Razavi3,4, David Sontag1. 1. CSAIL & IMES, Massachusetts Institute of Technology, Cambridge, MA. 2. Harvard Medical School, Boston, MA. 3. Memorial Sloan Kettering Cancer Center, New York, NY. 4. Weill-Cornell Medical College, New York, NY.
Abstract
PURPOSE: Key oncology end points are not routinely encoded into electronic medical records (EMRs). We assessed whether natural language processing (NLP) can abstract treatment discontinuation rationale from unstructured EMR notes to estimate toxicity incidence and progression-free survival (PFS). METHODS: We constructed a retrospective cohort of 6,115 patients with early-stage and 701 patients with metastatic breast cancer initiating care at Memorial Sloan Kettering Cancer Center from 2008 to 2019. Each cohort was divided into training (70%), validation (15%), and test (15%) subsets. Human abstractors identified the clinical rationale associated with treatment discontinuation events. Concatenated EMR notes were used to train high-dimensional logistic regression and convolutional neural network models. Kaplan-Meier analyses were used to compare toxicity incidence and PFS estimated by our NLP models to estimates generated by manual labeling and time-to-treatment discontinuation (TTD). RESULTS: Our best high-dimensional logistic regression models identified toxicity events in early-stage patients with an area under the curve of the receiver-operator characteristic of 0.857 ± 0.014 (standard deviation) and progression events in metastatic patients with an area under the curve of 0.752 ± 0.027 (standard deviation). NLP-extracted toxicity incidence and PFS curves were not significantly different from manually extracted curves (P = .95 and P = .67, respectively). By contrast, TTD overestimated toxicity in early-stage patients (P < .001) and underestimated PFS in metastatic patients (P < .001). Additionally, we tested an extrapolation approach in which 20% of the metastatic cohort were labeled manually, and NLP algorithms were used to abstract the remaining 80%. This extrapolated outcomes approach resolved PFS differences between receptor subtypes (P < .001 for hormone receptor+/human epidermal growth factor receptor 2- v human epidermal growth factor receptor 2+ v triple-negative) that could not be resolved with TTD. CONCLUSION: NLP models are capable of abstracting treatment discontinuation rationale with minimal manual labeling.
PURPOSE: Key oncology end points are not routinely encoded into electronic medical records (EMRs). We assessed whether natural language processing (NLP) can abstract treatment discontinuation rationale from unstructured EMR notes to estimate toxicity incidence and progression-free survival (PFS). METHODS: We constructed a retrospective cohort of 6,115 patients with early-stage and 701 patients with metastatic breast cancer initiating care at Memorial Sloan Kettering Cancer Center from 2008 to 2019. Each cohort was divided into training (70%), validation (15%), and test (15%) subsets. Human abstractors identified the clinical rationale associated with treatment discontinuation events. Concatenated EMR notes were used to train high-dimensional logistic regression and convolutional neural network models. Kaplan-Meier analyses were used to compare toxicity incidence and PFS estimated by our NLP models to estimates generated by manual labeling and time-to-treatment discontinuation (TTD). RESULTS: Our best high-dimensional logistic regression models identified toxicity events in early-stage patients with an area under the curve of the receiver-operator characteristic of 0.857 ± 0.014 (standard deviation) and progression events in metastatic patients with an area under the curve of 0.752 ± 0.027 (standard deviation). NLP-extracted toxicity incidence and PFS curves were not significantly different from manually extracted curves (P = .95 and P = .67, respectively). By contrast, TTD overestimated toxicity in early-stage patients (P < .001) and underestimated PFS in metastatic patients (P < .001). Additionally, we tested an extrapolation approach in which 20% of the metastatic cohort were labeled manually, and NLP algorithms were used to abstract the remaining 80%. This extrapolated outcomes approach resolved PFS differences between receptor subtypes (P < .001 for hormone receptor+/human epidermal growth factor receptor 2- v human epidermal growth factor receptor 2+ v triple-negative) that could not be resolved with TTD. CONCLUSION: NLP models are capable of abstracting treatment discontinuation rationale with minimal manual labeling.
Authors: Kenneth L Kehl; Wenxin Xu; Eva Lepisto; Haitham Elmarakeby; Michael J Hassett; Eliezer M Van Allen; Bruce E Johnson; Deborah Schrag Journal: JCO Clin Cancer Inform Date: 2020-08
Authors: Adam Yala; Regina Barzilay; Laura Salama; Molly Griffin; Grace Sollender; Aditya Bardia; Constance Lehman; Julliette M Buckley; Suzanne B Coopey; Fernanda Polubriaginof; Judy E Garber; Barbara L Smith; Michele A Gadd; Michelle C Specht; Thomas M Gudewicz; Anthony J Guidi; Alphonse Taghian; Kevin S Hughes Journal: Breast Cancer Res Treat Date: 2016-11-08 Impact factor: 4.872
Authors: Kenneth L Kehl; Haitham Elmarakeby; Mizuki Nishino; Eliezer M Van Allen; Eva M Lepisto; Michael J Hassett; Bruce E Johnson; Deborah Schrag Journal: JAMA Oncol Date: 2019-10-01 Impact factor: 31.777
Authors: Po-Ru Loh; George Tucker; Brendan K Bulik-Sullivan; Bjarni J Vilhjálmsson; Hilary K Finucane; Rany M Salem; Daniel I Chasman; Paul M Ridker; Benjamin M Neale; Bonnie Berger; Nick Patterson; Alkes L Price Journal: Nat Genet Date: 2015-02-02 Impact factor: 38.330
Authors: Kun-Hsing Yu; Ce Zhang; Gerald J Berry; Russ B Altman; Christopher Ré; Daniel L Rubin; Michael Snyder Journal: Nat Commun Date: 2016-08-16 Impact factor: 14.919