Cholera Antibacterial Resistance in Bangladesh: big data mining and machine learning to improve diagnostics and treatment selection

Burden

Cholera is a deadly disease with approximately 3-5 million cases and over 1,00,000 deaths annually worldwide. Of the 1.3 billion people at risk worldwide, 66 million are in Bangladesh equating to approximately 40% of the Bangladeshi population. In addition, refugee movement bring increased risk from this disease. Bangladesh is one of the Least Developed Countries list of ODA recipients and together with India has the largest population at risk of Cholera. Rapid diagnosis and early detection of outbreaks are key aspects to fight cholera. Moreover, the indiscriminate use of wide-spectrum antibiotics creates the additional threat of antibacterial resistance (ABR) in V. cholerae population.

Knowledge gap

Microbiological testing is resource-intensive, and outbreak detection is mostly based on unreliable reports of cholera-like diarrhoea cases from local hospitals. Advances in diagnostics, treatment selection and outbreak tracking are much-needed for progressing towards eliminating cholera as a public health threat by 2030, a recently proclaimed objective by the WHO-backed Global Taskforce for Cholera Control.

Relevance

The aggregation of geo-localised clinical, environmental, and societal information collected for the development of the diagnostic and early prediction systems, and the additional data continuously collected during the deployment and operation of such systems, will constitute an invaluable databank shareable across follow-on and collaborative projects and eventually across countries.

Hypothesis

Significant changes in understanding transmission dynamics of antimicrobial resistant V. cholerae in Bangladesh by big data mining and machine learning with better local community decision making to improve diagnosis and treatment of cholera.

Objectives

The specific objectives of this project are as follows:

• Develop a portable, real-time diagnostics solution for cholera infection caused by antimicrobial resistant V. cholerae
• Develop a surveillance system to identify hot spots with higher likelihood of outbreak
• Development of a shareable databank

Methods

Samples will be collected from Dhaka Hospital, Mathbaria Thana Health Complex, Cox’s Bazar Hospital and Rohingya camp. Immediate after collection, samples will be subjected to RDT. If the sample is positive for either V. cholerae O1 or O139 then one aliquot will be stored at -80°C freezer at icddr,b for future use and another aliquot will be transferred to NSU, Bangladesh for further analysis (Alam et al. 2006a). Water samples will also be collected from 6 sites each, for Dhaka city, Mathbaria, and Cox’s Bazar. Toxigenic V. cholerae will be isolated from stool and water samples following standard culture methods, and characterized for antibiotic resistance (Alam et al. 2006b, c). Both types of samples will be subjected to Nanopore genome sequencing.

Outcome measures/variables

Through the collaboration this proposal brings expertise together to work on public health. This will enable a much-needed multidisciplinary research programme to diagnose cholera using Nanopore genome sequencing, treatment selection, epidemiological forecasting for infection and antibacterial resistance, ultimately contributing to improving health, welfare and economic growth of Bangladesh.

Data mining and machine learning appear to offer better resolution for improving accuracy of diagnosis of a pathogen. The portable real-time nannopore sequencing device could provide diagnostic solution at field level. We designed this big data mining and machine learning study to improve diagnostics and treatment selection for cholera infection caused by antimicrobial resistant V. cholerae.

A portable real-time diagnostics solution for cholera infection caused by antimicrobial resistant V. cholerae with big data mining and machine learning to improve diagnostics and treatment selection.