AI detection of M. Tuberculosis pathogens using Generative Adversarial Network (GAN) analyses

Background

Rapid identification of pathogens is critical to outbreak detection and sentinel surveillance; however most diagnoses are made in laboratory settings. Advancements in artificial intelligence (AI) and computer vision offer unprecedented opportunities to facilitate detection and reduce response time in field settings. An initial step is the creation of analysis algorithms for offline mobile computing applications.

Methods

AI models to identify objects using computer vision are typically “trained” on previously labeled images. The scarcity of labeled image-libraries creates a bottleneck, requiring thousands of labor hours to annotate images by hand to create “training data.” We describe the applicability of Generative Adversarial Network (GAN) methods to amass sufficient training data with minimal manual input.

Results

Our AI models are built with a performance score of 0.84-0.93 for M. Tuberculosis, a measure of the AI model's accuracy using precision and recall. Our results demonstrate that our GAN pipeline boosts model robustness and learnability of sparse open source data.

Conclusions

The use of labeled training data to identify M. Tuberculosis developed using our GAN pipeline techniques demonstrates the potential for rapid identification of known pathogens in field settings. Our work paves the way for the development of offline mobile computing applications to identify pathogens outside of a laboratory setting. Advancements in artificial intelligence (AI) and computer vision offer unprecedented opportunities to decrease detection time in field settings by combining these technologies. Further development of these capabilities can improve time-to-detection and outbreak response significantly.

Key messages

Rapidly deploy AI detectors to aid in disease outbreak and surveillance. Our concept aligns with deploying responsive alerting capabilities to address dynamic threats in low resource, offline computing environs.

Does dioxin exert toxic effects in humans at or near current background body levels?: an evidence-based conclusion

Evidence-based toxicology like evidence-based medicine, provides scientifically grounded evidence-based conclusions as distinguished from authority-based opinions. As an example, we address a proposition from the US Environmental Protection Agency's (EPA) Draft Dioxin [2,3, 7,8-tetrachlorodibenzo-P-dioxin (TCDD)] reassessment that: ‘dioxin... can produce effects... at or near current background body burdens or intake levels’. Guided by a systematic, objective, and unbiased analysis of the available molecular, physiological, and clinical/epidemiologic data, in accordance with accepted principles of scientific logic, we reach the evidence-based conclusion that the proposition is rejected. When gaps in scientific knowledge necessitate formulation of opinions to meet preventive or precautionary goals, the reversion to authority should be explicitly acknowledged.

Structural and functional characterization of liver cell-specific activity of the human sodium/taurocholate cotransporter

Bile salts are rapidly removed from the circulation by the liver-specific sodium/taurocholate cotransporter (SLC10A1). To understand factors controlling its liver-specific expression, we isolated human SLC10A1 from a YAC chromosomal clone. SLC10A1 spans approximately 23 kb distributed over five exons. The major transcription start site is at 299 bp, and a minor start site is at 395 bp from the translational start site. A 1.2-kb portion of the 5' flanking region was sequenced and shown to contain a number of liver-enriched elements, but no TATA box. Using secreted alkaline phosphatase reporter constructs liver-specific expression was examined. Transient transfection demonstrated that SLC10A1 promoter expression was selectively expressed eightfold in FAO and rat hepatocytes, while deletion mutants demonstrated liver-specific expression in a region extending from -5 to +198 bp, which contained putative sites for C/EBP and HNF3. Mutations of the C/EBP site resulted in loss of 77% of transcriptional activity. Cotransfection of C/EBP, but not other putative liver-enriched binding factors, increased SLC10A1 promoter activity. Electrophoretic mobility shift assays demonstrated specific protein-DNA interactions that involved C/EBPalpha and beta. These studies demonstrate that the TATA-less human SLC10A1 promoter exhibits liver-specific activity and its regulatory elements contain binding sites for C/EBP, which contributes specifically to its transcriptional regulation.

Mechanism for diminished tissue factor expression by endothelial cells cultured with heparin binding growth factor-1 and heparin

We have extended our earlier observation that growing primary cultures of human umbilical vein endothelial cells (HUVEC) with heparin binding growth factor 1 (HBGF-1) 20 micrograms/mL and heparin 12 U/mL inhibits expression of tissue factor (TF) activity on HUVC monolayers perturbed with thrombin. TF activity was measured as the ability of monolayers or cell lysates to support FVIIa-catalyzed activation peptide release from 3H-FX. TF antigen in HUVEC extracts was measured in an enzyme-linked immunosorbent assay (ELISA) that uses a double-antibody sandwich technique with rabbit and goat antibodies to human TF. TF-mRNA was measured by Northern blot hybridization with a 32P-TF cDNA probe. Cells growth with HBGF-1/heparin had both decreased surface and total TF activity as compared with HUVEC from the same endothelial cell pool grown without HBGF-1/heparin. Means +/- SD for TF antigen for four primary cultures were 4.4 +/- 0.9 ng/10(6) cells without HBGF-1/heparin and 0.6 +/- 0.3 ng/10(6) cells with HBGF-1/heparin. TF mRNA 4 hours after incubation with thrombin of HUVEC grown without HBGF-1/heparin was about sevenfold higher than TF mRNA of HUVEC grown with HBGF-1/heparin. These data establish that growing primary cultures of HUVEC with HBGF-1/heparin impairs their ability to synthesize TF apoprotein after perturbation. This may be part of a generalized response of endothelial cells to HBGF-1/heparin facilitating migration during angiogenesis.