Workshop on Health Analytics and Disease Modelling
Organized by National Disease Modelling Consortium (NDMC), IIT Bombay

February 23-24, 2023

The workshop aimed to train budding mathematical/statistical modelers to enable a better understanding of the key concepts in health data analytics, machine-learning (ML) methods in healthcare, spatial disease mapping, economic and epidemiological models. Hands-on training sessions in Python on ML, semi-supervised ML methods, SIR and SEIR models, and hotspot mapping were conducted as part of the workshop. The focus of the workshop was on Neglected Tropical Diseases (NTDs), including lymphatic filariasis (LF) and visceral leishmaniasis (VL; kala-azar), which cause high morbidity and mortality in low and middle-income countries. The disease burden of NTDs and programmatic updates were also discussed in the workshop in the form of presentations by various faculty members and eminent speakers.

More than 80 participants attended the workshop. Travel grants were awarded to 9 people. Participants were mainly from IIT Bombay, IIT Kanpur, Indian Statistical Institute Kolkata, University of Hyderabad, ICAR NiVeDi, IIPS, Kokilaben Dhirubhai Ambani Hospital and Medical Research Institute Mumbai, Maharashtra University of Health Sciences (MUHS) Nashik, UNICEF Raipur and other institutes like NIRRH, Chennai Mathematical Institute (CMI), and IISc Bangalore.


On Day 1, Professor Siuli Mukhopadhyay from the Department of Mathematics at IIT Bombay and principal investigator of NDMC started the workshop with a brief introduction of NDMC and a presentation on Statistics and Data Analysis in Public Health. Basic concepts in epidemiology such as measures of incidence, prevalence and mortality; epidemiological studies including cross-sectional, case-control and cohort studies; as well as the relative risk and odds-ratio were discussed. This introductory session on epidemiology generated great interest in the audience given only few were statisticians. Furthermore, these basic concepts provided context for more advanced and applied statistical concepts in health that were presented at the latter part of the presentation. The concepts discussed included probability distributions and their estimation as well as hypothesis testing using examples from various disease states. A comment from the post workshop feedback survey suggested including additional examples of statistical analysis using real-time data. This will be considered for upcoming workshops.

Professor Harish Phuleria, from the Department of Environmental Science and Engineering at IIT Bombay delivered a presentation on Spatial Analysis for Disease Data and Disease Mapping. The utilization of disease surveillance systems for data collection, and analysis and interpretation of these data to provide strategies for disease control and address other public health objectives were discussed. Selection of methods for space-time disease surveillance was outlined. Other topics such as the advantages of disease mapping, prioritization of diseases for mapping, and examples of global mapping of diseases were presented. The group's recent research on the spatiotemporal distribution of malaria and dengue in India was discussed. The audience had several comments. In response to Professor Harish’s concern that access to health data has been a challenge for researchers, participants from Maharashtra University of Health Sciences and Indian Council of Medical Research proposed collaborating and sharing their disease related data with the NDMC team at IIT Bombay for modelling purposes. Given the consistent efforts by Indian researchers to understand and treat people with NTDs with the ultimate aim to eliminate these diseases, the audience questioned the validity of the term “neglected”—a term coined by the World Health Organization—in the Indian context.

The next presentation by Professor Mithun Mitra from the Department of Physics at IIT Bombay focused on Compartmental Models in Epidemiology. Basic mathematical models applied to vector-borne diseases including compartmental, spatial, and agent based models were introduced followed by a detailed discussion on the Kermack-McKendrick epidemic (or SIR) model for the spread of disease as well as the vaccination model. Additionally, a recent article on modelling the epidemiology and control of VL or kala azar, one of the common NTDs in India, was discussed. The importance of modelling different interventions to guide public policy was highlighted. For example, according to the article, vector-related measures were more effective than the treatment-related measures for the control of kala azar in India. In this stimulating session, the audience enthusiastically participated in proposing potential assumptions to be considered when generating an SIR model and raised several other interesting questions.

The final presentation of Day 1 was delivered by Professor Souvik Banerjee, from the Department of Economics at IIT Bombay and principal investigator of the NTD group, on the Methods for Health Economic Evaluation. In this engaging presentation, efficiency considerations and its need with a focus on the healthcare sector was discussed. The presentation covered details of the steps involved in an economic analysis that included defining a problem, stating objectives and identifying alternatives, constructing a decision tree, and analysing benefits/effects and costs. Finally, economic evaluations such as cost-benefit, cost-effectiveness and cost-utility analysis were discussed using examples of medical treatments. As only a few people in the audience had a background in economics, the topic generated keen interest and led to several one-on-one discussions.

Day 2 started with an informative talk by Dr. Bhupendra Tripathi, Country Lead - Elimination Programs, Neglected Tropical Diseases of the Bill & Melinda Gates Foundation on Neglected Tropical Diseases - Disease Burden and Programmatic Updates. A background on NTDs in India, India’s global ranking for common NTDs, and the global elimination targets for 2030 were presented. Insights on disease characteristics, diagnosis, transmission and elimination programmes for LF and VL were provided. Since the elimination of LF by 2027 is a public health priority for the Government of India, the importance of implementation of mass drug administration and drug consumption patterns were highlighted. Most importantly, modelling questions that need to be addressed including the time required for elimination, the effect of people migration and the untreated on disease transmission, impact of vector density on transmission and identification of transmission hotspots were presented.

The final presentation on Telemedicine technologies for NTD management was delivered by Professor Nirmal Punjabi from the Koita Center for Digital Health at IIT Bombay. The presentation highlighted the potential of these technologies, especially mHealth, in facilitating support to the healthcare worker with tools to help with screening, detection, monitoring the progression and treatment efficiency, to alleviate the burden of NTDs. The presentation provided insights into the telemedicine framework and its critical building block. mHealth was introduced as a potential solution for complete telemedicine management in remote settings. Various sensors and communication modules as an integral part of smartphones, and their role for disease surveillance were discussed. Furthermore, the basics of LF and VL from the perspective of diagnosis and monitoring and various apps used for Skin NTD management were discussed. Technical impediments to implementing these technologies, such as connectivity issues, the need for specialized devices and software, and data privacy and security concerns and solutions offered to address these challenges were discussed.

Python Sessions

Hands-on Python sessions were conducted on each day based on the lectures delivered. On Day 1, the session on Disease Mapping and Epidemiological Models was conducted. This session focussed on introducing different types of maps and disease compartmental models. Plotting of boundaries of states of India was discussed. Plotting of graduated symbol map, choropleth, both individually and together were taught. SIR model with births and deaths, and SVIR model were also discussed. Initialization of values of various compartments at time 0, coding and solving of differential equations for the above mentioned models, plotting of values of different compartments and identification of equilibrium time was also taught in this session. Also, some exercise questions were given to participants.

On Day 2, a session on Introduction to ML Models & Hands-on Python Sessions on ML Methods was conducted. This session focused on introduction to Python for those who were using Python, Numpy for the first time. This was followed by estimation and modeling of linear regression. Perceptron classification of tabular data, and image data. The MNIST database (Modified National Institute of Standards and Technology database), which is a large database of handwritten digits that is commonly used for training various image processing systems, such as PathMNIST, PneumoniaMNIST and OrganAMNIST, was also taught during the session. How to perform a subset selection for research work from a huge population using machine learning techniques, was also discussed.