New adjunct specialty facilities in India's leading health institutions are advancing precision medicine in the country through AI. 

The Apollo Cancer Centre (ACC) in Bengaluru has recently opened what could be the country's first Precision Oncology Centre powered by AI. It offers comprehensive, specialised care that is tailored to each individual. 

The centre features AI automation to identify eligible patients for targeted therapy and immunotherapy, as well as to alert care teams of patient deterioration. It also deploys conversational AI to educate patients and their families on diagnosis, treatment, and connections to support groups. Additionally, it utilises AI to monitor adherence to standard care; enable patient management based on genomic, clinical, and pathological data; and offer recommendations for diagnostic tests and enrolment to value-based care and other patient benefit programmes.

ACC touts that the use of AI, along with harnessing volumes of health data, is the "future of oncology." With AI, it can ensure accurate diagnosis, real-time insights, cancer risk assessment, treatment protocol and continuum of care. 

Leveraging its expertise in AI, the Indian Institute of Science (IISc), meanwhile, recently launched a collaborative laboratory for AI in Precision Medicine with Siemens Healthineers, a known brand in medical imaging. The lab will develop open-source AI-based tools to automatically segment pathological findings in brain scans. Soon to be integrated into regular clinical workflows, these tools are intended to assist in accurately diagnosing neurological diseases and analysing their clinical impact at a population level. 

Drs Vijay Agarwal and Vishwanath S, senior consultants of Medical Oncology at ACC, shared with Healthcare IT News more details about their applications of AI in precision oncology. Vaanathi Sundaresan, assistant professor at IISc Department of Computational and Data Sciences and head of the Siemens Healthineers-Computational Data Sciences Collaborative Laboratory for AI in Precision Medicine discussed how they intend to protect sensitive patient data amid growing cybersecurity threats. 

Q: Can you share specific use cases or applications of AI in your new facility?

Dr Agarwal, ACC: There's this one case: a woman with a lump in her breast who visited us for a consultation. Using AI, she was immediately diagnosed with breast cancer within 24 hours of presentation. Following the diagnosis, an automated alert notified all stakeholders – the treating physician, the lead breast surgeon, the multi-disciplinary team (MDT) coordinator and the patient – about the need for an MDT. Once the MDT meeting was held, a recommendation was sent to the centre, and then treatment commenced. The patient pathway was predefined using AI, and all stakeholders were made aware of it. Once the treatment, which includes chemotherapy, was planned, auto alerts were built in for a seamless process of admission, drug ordering, chem prescribing, drug delivery, consents (specific to drug regimens and language), discharge, and payments, thereby improving efficiency and reducing costs. Every change in the treatment plan was relayed automatically to all stakeholders, thereby making the care seamless and well-integrated across all specialties. Chemotherapy and targeted therapy were later advised and then the patient was eventually referred to the MDT. 

Dr Vishwanath S, ACC: We use AI that facilitates early, seamless delivery of chemotherapy right from registration, bed booking to discharge. AI also plays a role in facilitating personalised therapy based on NGS (next-generation sequencing) mutation status. In addition, digital pathology and images can be AI-driven – a good example is using bioinformatics and AI to identify a patient with heavily pre-treated advanced sarcoma and an NGS report showing a targetable mutation.

A/Prof Sundaresan, IISc: Some other relevant applications of AI that are quite important for clinical deployment include population-level modelling of disease progression, adapting the AI models to be robust towards variation in data characteristics across sites (domain adaptation), limited availability of data (low-data regimes), scarcity of manual labels and outliers. Another important long-term direction would be to identify the relationship between brain health and other organs of the body.

Q: What are you planning to be the first project of the collaborative lab? How urgent is the need for precise imaging/diagnosis of neurological diseases and how does AI can support this?

A/Prof Sundaresan: Our first project will be the identification of vascular biomarkers on neuroimaging data that would aid in the early detection of neurodegeneration.

The prevalence of neurodegenerative diseases (such as Alzheimer's disease and other types of dementia) and cerebrovascular diseases like stroke have been associated with cognitive impairment, gait disturbances, and brain atrophy, which at times can lead to death (with fatality rates up to 47% reported for stroke) and commonly found in subjects with vascular risk factors and depression. AI methods applied to MRI scans can lead to the detection of imaging biomarkers for personalised treatment. However, differential diagnosis and long-term prognosis of such neurological diseases require highly specific imaging biomarkers and thorough investigation of their precise clinical impact – and this is where AI methods can be quite useful. 

Q: Given the lab's extensive use of sensitive data, how do you intend to secure and protect this data and the algorithms/models that you will apply? 

A/Prof Sundaresan: Most of the experiments used in the lab will involve publicly available data for initial testing. Any clinical data acquired for the research (from IISc or collaborators) will be obtained after ethical board clearances and will be strictly anonymised and privacy preserved. The methods (without training data) will be open-source for the benefit of the wider research community.

_

Their responses have been edited for brevity and clarity.