Machine learning classification of surgical pathology reports and chunk recognition for information extraction noise reduction

Abstract

Napolitano, Giulio et al., Machine learning classification of surgical pathology reports and chunk recognition for information extraction noise reduction, Artificial Intelligence in Medicine, 2016 Machine learning classification of surgical pathology reports and chunk recognition for information extraction noise reduction Giulio Napolitano Materials and methods:The first technique was implemented using the freely available software RapidMiner to classify the reports according to their general layout: 'semi-structured' and 'unstructured'. The second technique was developed using the open source language engineering framework GATE and aimed at the prediction of chunks of the report text containing information pertaining to the cancer morphology, the tumour size, its hormone receptor status and the number of positive nodes. The classifiers were trained and tested respectively on sets of 635 and 163 manually classified or annotated reports, from the Northern Ireland Cancer Registry. Results:The best result of 99.4% accuracy -which included only one semi-structured report predicted as unstructured -was produced by the layout classifier with the k nearest algorithm, using the binary term occurrence word vector type with stopword filter and pruning. For chunk recognition, the best results were found using the PAUM algorithm with the same parameters for all cases, except for the prediction of chunks containing cancer morphology. For semi-structured reports the performance ranged from 0.97 to 0.94 and from 0.92 to 0.83 in precision and recall, while for unstructured reports performance ranged from 0.91 to 0.64 and from 0.68 to 0.41 in precision and recall. Poor results were found when the classifier was trained on semi-structured reports but tested on unstructured. Conclusions: These results show that it is possible and beneficial to predict the layout of reports and that the accuracy of prediction of which segments of a report may contain certain information is sensitive to the report layout and the type of information sought.

Applications of Machine Learning Predictive Models in the Chronic Disease Diagnosis

“…With ML models, it can also be possible to improve quality of medical data, reduce fluctuations in patient rates, and save in medical costs. Therefore, these models are frequently used to investigate diagnostic analysis when compared with other conventional methods [10]. To reduce the death rates caused by chronic diseases (CDs), early detection and effective treatments are the only solutions [11].…”

Obtaining Knowledge in Pathology Reports Through a Natural Language Processing Approach With Classification, Named-Entity Recognition, and Relation-Extraction Heuristics

“…Despite near-universal electronic medical record use, pathology reports remain as free text containing semistructured elements detailing a specimen's source and gross and microscopic characteristics. Although other groups have developed tools to aid in parsing the cancer type or tumor characteristics from these reports, such as to identify relevant patients for registry inclusion [1][2][3][4] or to determine TNM staging, [5][6][7][8][9] none, to our knowledge, have attempted to extract and group semistructured specimen identifiers themselves.…”

Next Generation Quality: Assessing the Physician in Clinical History Completeness and Diagnostic Interpretations Using Funnel Plots and Normalized Deviations Plots in 3,854 Prostate Biopsies

“…The accuracy of free text interpretations in pathology varies substantially; it can be nearly perfect (99% accurate) or be quite poor (65%). [ 2 ] Seen practically, in the context of analyzing free-text pathology reports, this may limit analysis work on conditions that have a prevalence lower than the error rate.…”