Publications

BACKGROUND

An accurate system for tracking of colonoscopy quality and surveillance intervals could improve the effectiveness and cost-effectiveness of colorectal cancer (CRC) screening and surveillance. The purpose of this study was to create and test such a system across multiple institutions utilizing natural language processing (NLP).

METHODS

From 42,569 colonoscopies with pathology records from 13 centers, we randomly sampled 750 paired reports. We trained (n=250) and tested (n=500) an NLP-based program with 19 measurements that encompass colonoscopy quality measures and surveillance interval determination, using blinded, paired, annotated expert manual review as the reference standard. The remaining 41,819 nonannotated documents were processed through the NLP system without manual review to assess performance consistency. The primary outcome was system accuracy across the 19 measures.

RESULTS

A total of 176 (23.5%) documents with 252 (1.8%) discrepant content points resulted from paired annotation. Error rate within the 500 test documents was 31.2% for NLP and 25.4% for the paired annotators (P=0.001). At the content point level within the test set, the error rate was 3.5% for NLP and 1.9% for the paired annotators (P=0.04). When eight vaguely worded documents were removed, 125 of 492 (25.4%) were incorrect by NLP and 104 of 492 (21.1%) by the initial annotator (P=0.07). Rates of pathologic findings calculated from NLP were similar to those calculated by annotation for the majority of measurements. Test set accuracy was 99.6% for CRC, 95% for advanced adenoma, 94.6% for nonadvanced adenoma, 99.8% for advanced sessile serrated polyps, 99.2% for nonadvanced sessile serrated polyps, 96.8% for large hyperplastic polyps, and 96.0% for small hyperplastic polyps. Lesion location showed high accuracy (87.0-99.8%). Accuracy for number of adenomas was 92%.

CONCLUSIONS

NLP can accurately report adenoma detection rate and the components for determining guideline-adherent colonoscopy surveillance intervals across multiple sites that utilize different methods for reporting colonoscopy findings.

MicroRNAs (miRNAs) are small non-coding RNAs that post-transcriptionally control the expression of their target genes via RNA interference. There is increasing evidence that expression of miRNAs is dysregulated in neuronal disorders, including epilepsy, a chronic neurological disorder characterized by spontaneous recurrent seizures. Mesial temporal lobe epilepsy (MTLE) is a common type of focal epilepsy in which disease-induced abnormalities of hippocampal neurogenesis in the subgranular zone as well as gliosis and neuronal cell loss in the cornu ammonis area are reported. We hypothesized that in MTLE altered miRNA-mediated regulation of target genes could be involved in hippocampal cell remodeling. A miRNA screen was performed in hippocampal focal and non-focal brain tissue samples obtained from the temporal neocortex (both n=8) of MTLE patients. Out of 215 detected miRNAs, two were differentially expressed (hsa-miR-34c-5p: mean increase of 5.7 fold (p=0.014), hsa-miR-212-3p: mean decrease of 76.9% (p=0.0014)). After in-silico target gene analysis and filtering, reporter gene assays confirmed RNA interference for hsa-miR-34c-5p with 3'-UTR sequences of GABRA3, GRM7 and GABBR2 and for hsa-miR-212-3p with 3'-UTR sequences of SOX11, MECP2, ADCY1 and ABCG2. Reporter gene assays with mutated 3'-UTR sequences of the transcription factor SOX11 identified two different binding sites for hsa-miR-212-3p and its primary transcript partner hsa-miR-132-3p. Additionally, there was an inverse time-dependent expression of Sox11 and miR-212-3p as well as miR-132-3p in rat neonatal cortical neurons. Transfection of neurons with anti-miRs for miR-212-3p and miR-132-3p suggest that both miRNAs work synergistically to control Sox11 expression. Taken together, these results suggest that differential miRNA expression in neurons could contribute to an altered function of the transcription factor SOX11 and other genes in the setting of epilepsy, resulting not only in impaired neural differentiation, but also in imbalanced neuronal excitability and accelerated drug export.

Drug discovery and development are hampered by high failure rates attributed to the reliance on non-human animal models employed during safety and efficacy testing. A fundamental problem in this inefficient process is that non-human animal models cannot adequately represent human biology. Thus, there is an urgent need for high-content in vitro systems that can better predict drug-induced toxicity. Systems that predict cardiotoxicity are of uppermost significance, as approximately one third of safety-based pharmaceutical withdrawals are due to cardiotoxicty. Here, we present a cardiac microphysiological system (MPS) with the attributes required for an ideal in vitro system to predict cardiotoxicity: i) cells with a human genetic background; ii) physiologically relevant tissue structure (e.g. aligned cells); iii) computationally predictable perfusion mimicking human vasculature; and, iv) multiple modes of analysis (e.g. biological, electrophysiological, and physiological). Our MPS is able to keep human induced pluripotent stem cell derived cardiac tissue viable and functional over multiple weeks. Pharmacological studies using the cardiac MPS show half maximal inhibitory/effective concentration values (IC₅₀/EC₅₀) that are more consistent with the data on tissue scale references compared to cellular scale studies. We anticipate the widespread adoption of MPSs for drug screening and disease modeling.