Machine learning/artificial intelligence for healthcare

Artificial intelligence (AI) in healthcare is typically the use of software and complex machine learning algorithms with self-learning neural networks to gather, process, interpret, and present medical and healthcare data mimicking human cognition. AI has been helpful in the areas of drug development and interactions, electronic health records, disease diagnosis, primary care, telemedicine, radiology, dermatology etc. Faculties from the department of chemistry, data science and engineering, electrical engineering and computer science are interested to take-up research under machine learning/artificial intelligence area. Some of the research problems relevant to the area are as under:

a. Biased signalling in G-protein-coupled receptors:

GPCRs are primary targets for more than one-third of the marketed drugs. However, drugs are generally equally potent in activating both the G-protein and b-arrestin pathways, thereby unable to separate therapeutic effects from adverse side effects. Over the past decades, though the ligand-dependent differential efficacy for distinct intracellular responses through the same receptor has been well-documented for many GPCRs, but for the most part both structural as well as dynamical basis for this selectivity are poorly understood. Thus, obtaining a detailed insight into the molecular signatures of biased signalling is therefore essential to design and develop novel therapeutic molecules for the effective clinical treatment of various diseases. Systematic investigation of the specific changes in the receptor domains that are caused by biased ligands and/or mutants, and the differences in the manner in which they elicit the underlying activation of that receptor is a key priority.

Internet of Things (IoT) for healthcare

The remote monitoring of patients has been made possible through Internet of Things (IoT). IoT has empowered medical practitioners to perform superlative care and proactively connect with their patients. The overall logic is to interconnect various devices with sensors, monitors, detectors, cameras, for data collection followed by the digitization of collected data and advanced analysis. This results in better healthcare, reduced treatment costs, improved process and performance for healthcare providers. Faculties from the department of electrical engineering and computer sciences and data science and engineering are interested to conduct research in this area.

Some of the research interests are development of IoT and IT based healthcare solutions, bioelectronic interfaces, and futuristic devices. a. Clinical question answering and dialogue generation: Currently, most of the tasks and enquiries at the hospital reception are resolved by the humans. This limits the task resolving rate and increases the operational cost. With the help of a chatbot model, a digital platform can be developed to resolve the tasks and queries without human intervention. A neural model can be trained on a publicly available english medical conversations dataset to resolve relatively simpler tasks. Its capabilities can be improved to handle complex tasks by further training the model on the recorded actions of the hospital staff. Use of contextual knowledge bases with transformer models to build the model. This will make the model better and 9 efficient over time. Various pre-trained medical language models like Clinical-BERT, Bio-BERT, etc. can be used to initialize the weights of the query encoder. Over time, with the help of feedback from the doctors, the system can be improved for better performance.

Cancer biology research

Research on the understanding the mechanisms behind cell growth, transformation of normal cells into cancer cells, and study on spread of cancer cells fall under cancer biology research. Department of biology and chemistry are interested in this research area. Some of the priority research themes are study on cancer epigenetics and alternative splicing in breast and head & neck cancer, understanding the nuclear pore biology and posttranslational modification induced physiological changes leading to enhanced cellular proliferation and cancer, to understand the connection between nuclear pore proteins' overexpression and the progression/ aggressiveness of cancer, to extend nuclear pore-related studies to hematologic and bone disorders ensued by the malfunctioning of constituent cells of these tissues, proteomic study on the analysis of SUMO modification dynamics during apoptosis and autophagy.

Material synthesis for anti-cancer potential: FDA approved drugs containing gold, namely Auranofin, Solganol, Myocrisin, Sanochrysin and Allocrysine, have witnessed enormous success in treating rheumatoid arthritis due to their immunosuppressive properties. Recently, gold complexes have also found tremendous potential as anticancer agents and some are under clinical trials at various stages. Gold-based organic molecules are potential alternatives to the largely used platinum-based anticancer drugs. Mechanistically, unlike platinum complexes (for instance, cisplatin) that target DNA, gold complexes exhibit a high propensity to target enzymes, especially those containing thiols, selenothiol, etc. This distinct targeting feature is attributed to the strong binding affinity of gold towards thiol (soft ligands) containing enzymes, such as TrxR, glutathione reductase (GR), and cysteine protease, which are overexpressed in cancer cells. Notably, this complementarity between the anticancer mechanisms of gold- and platinum-based complexes should also allow the former to be efficient for the treatment of patients who developed resistance towards platinum-based drugs. So far, a relatively large number of gold(I) complexes have been synthesized and evaluated for their anticancer activities in the literature. Breast cancer cells 10 However, reports on the consolidated study that aims at the rational design of gold complexes and thorough evaluation of their anticancer properties remained scarce. There is a need to rationally design various gold complexes bearing diverse ligands and investigate their anti-cancer potential.

Disease and diagnosis markers for early diagnostics and visualization

Diagnosis markers are the biological parameters that help in disease diagnose. Some of the key examples are presence of genetic variant and the levels of certain proteins in body fluids. Department of biology, chemistry, data science and engineering are interested to pursue research on disease and diagnose markers. Some of the key problems of interest under disease and diagnosis markers are:

a. Biology of epigenetics and development of diagnosis markers for liver and poultry diseases:

The upcoming area of interest is to decipher the role of epigenetics in tissue specific gene expression more specifically in poultry liver during development and infections and to develop certain markers for the diagnosis of diseases for the management of poultry health. Through the utilization of budding yeast as a model organism, identifying the significance of epigenetics and conserved cellular homeostatic pathways in response to promising anti-cancer molecules, metals and environmental pollutants is a research priority.

b. Rationally designed small-molecule probe for liver and kidney damage:

Accurate analysis of liver and kidney damage monitoring-biomarker is urgently required for chronically ill patients in a ‘point-of-care’ fashion. Lately, the number of kidney disease is exponentially increasing due to our fast lifestyle and food habit. The technology demanding ‘strip-based’ detection by exploiting the high sensitivity in solution as well as in solid-state has been determined. It provided a convenient and cost-effective analysis of the protein-based biomarker associated with kidney and liver damage. The abnormal concentration of the marker in body fluids can be a potential biomarker for liver and kidney malfunction.

c. Visualisation of sub-cellular process:

The recent advancement of fluorescence microscopy allows to investigate the cellular processes with high spatial and temporal resolution in all three dimensions. The discovery of genetically encoded fluorescent proteins (FPs) allows for endogenous labelling and makes almost any protein or peptide fluorescently visible inside live-cells. However, in many cases the poor photo-stability, variable maturation time of the chromophoric unit in FPs and bulkiness limit their applicability. Therefore, there is high demand for the development of novel fluoroprobe with desirable properties. Development of biocompatible fluorescent dyes for deciphering the inter-organelle communication with high resolution is the key priority.

d. Fluorescent biomarking for live-cell imaging:

Live-cell imaging of subcellular organelles such as mitochondria, lysosomes, endoplasmic reticulum etc. has drawn a great attraction to visualize various biological phenomena. Fluorescence imaging has an advantage over other technologies (radioisotope labelling, electrochemical detections, MRI) in terms of sensitivity, safe-detection, tuneable fluorescence signal etc. A superior bioimaging fluorescent probe should have (i) high photostability for long-time bioimaging of organelles, (ii) sufficiently high solubility in subcellular medium, and (iii) low toxicity and biocompatibility. Fluorescent heteroaromatic compounds would be ideal candidates for bioimaging molecules. Since a limited number of bioimaging agents are available, current research is highly intensified to develop bioimaging molecules cost-effectively.

e. Strategies for proliferation, separation, and imaging of pancreatic β cells:

Currently 425 million people worldwide are diagnosed with diabetes. Out of these ~95% suffer from type 2 diabetes and ~5% suffer from type 1 diabetes. According to world diabetic federation, in South East Asia, especially in Indian subcontinents, ~82 million people are diagnosed with diabetes. In next 20 years, it is expected to increase to ~151 million. Apparently, the death and dysfunction of insulin-producing β-cell in the pancreas is the key pathological event in diabetes. As an alternative to the whole islet transplantation to the diabetic patient, iPS cell derived β-cells formation is expected to be promising for the treatment of diabetic patients. One of the pioneers in iPS cell derived β-cell generation is Prof. Dough Melton at Harvard Department of Stem Cell and Regenerative Biology. They have successfully stabilized the protocol by sequential modification of different signalling pathways and generated millions of β-cells from human pluripotent stem cell. They have validated the formation of β- cell by comparing the gene expression pattern between the iPS cell derived β- cell and pancreatic β-cell from a donor. However, the major bottleneck of this process is the overall yield of β-cell production - the typical yield of β-cell formation is 1-5%. Furthermore, this method also produces other undesired endocrine cells and contains undifferentiated iPS cells which could be tumorigenic. There is no available method to isolate these β-cells from the other pancreatic cell types. Research interest is to develop various chemical tools to separate and purify the β cells which are derived from iPS cells. Further interest is developing chemical tools to increase the yield by selectively delivering proliferating molecules and developing fluorescent probes for β cell imaging.

Novel material synthesis for healthcare

a. Soft nanomaterials for biomedical applications. These include hydrogels for drug delivery, polymers as novel antimicrobials, underwater adhesives for biomedical applications, novel fluorophores for in cellulo imaging applications, ROS generating metal-peptide complexes in anticancer applications.

b. Functional nanocomposites: developing functional materials in the form of films, fibres, or hydrogels, hydrogel patches for transdermal drug delivery, smart hydrogels for wound dressing using photothermal therapy, hydrogel-based scaffolding polymeric biomaterials for tissue engineering, antimicrobial hydrogels for potential medical applications, injectable and biocompatible hydrogel for drug delivery and tissue engineering, smart materials for wearable devices and flexible bioelectronics.