Project Lead(s): Prakash Naidu
An estimated 300 million people have diabetes, with 80% of deaths from the disease occurring in low- and middle-income countries.
A major cause of this is due to insulin resistance (IR), which cannot be detected prior to the development of type 2 diabetes.
The current methods for diagnosing IR are expensive, complex and time-consuming, and are not used widely, even in the developed world.
In this proof-of-concept study, the project team sought to develop a diagnostic test for insulin resistance that is easy to interpret, low cost, and delivers the best performance characteristics when compared to the gold standard.
They explored a low-cost, preventive diagnostics approach, using spectroscopic analysis.
Spectral analysis techniques were researched and assessed, with blood serum samples from 22 subjects with and without diabetes.
Each sample was a collection of absorbance values for 3,600 wavenumbers and, for better effectiveness of the classification techniques, it was necessary to reduce the number of features to identify the discriminating wavenumbers in the data, using a combination of classifiers.
Three classifiers – linear discriminant analysis (LDA), artificial neural network (ANN) and support vector machine (SVM) – were identified and used in combination to increase accuracy and reduce the number of features, in order to identify the discriminating wavenumbers in the data.
As part of an education awareness program associated with the project, villagers were told about the need to see a physician before diabetes is advanced.
Proof of concept could only be achieved partially, due to time and resource constraints.The team tried to evolve alternate methods for low-cost spectroscopy but, halfway through the project, realised that the requirements for IR detection through spectroscopy cannot likely be fulfilled by lower-cost solutions, given the time and resource constraints of the Phase I project.
However, centralized, higher-cost, spectroscopy-based IR detection was experimented that established pathways towards likely potential for a business model involving more investment.
The project was conducted in India but additionally enabled networking with stakeholders in Sri Lanka and Bhutan.