A recent collaboration between ICMR and IBM intends to use virtual assistants while supercomputing, AI and data technologies are already being used as strong shots in other places. Let’s see how deep this needle goes.
There must be reason why medical think-tanks are picking some technologies over others. Some tools are better directed to the skin because the right cells there make sure that the body learns to fight the enemy in an embedded way. Some are more predisposed to a certain vein or muscle because that’s when speed matters most.
Indian Council of Medical Research (ICMR), Department of Health Research, Ministry of Health and Family Welfare, Government of India, has collaborated with IBM. This is for implementation of a Watson virtual agent (called Watson Assistant) on its portal for responding to specific queries of front line staff and data entry operators from various testing and diagnostic facilities across the country on COVID-19. The agent’s job description also includes support for on-boarding new data entry operators and staff of diagnostic centres, as the COVID-19 test network expands across the country.
When this was announced, Professor Balram Bhargava, Director General, Indian Council of Medical Research (ICMR) had underlined the need to augment teams' response time and allow them to concentrate on priorities like developing and updating testing and treatment protocols and guidance for COVID-19.
We ask Gargi Dasgupta—Director-IBM Research India and CTO-IBM India/South Asia to tell us more about the relevance, significance, scope and struggles that define the choices the healthcare community is making.
What spurred the idea of a virtual agent—what exact value and role will this bring in the exploration effort?
The pandemic has led to a change in social norms including everything from reduced physical interaction and contact; to an explosion in digital communication and services. It is difficult for professionals such as doctors, teachers, lawyers, etc. to handle the deluge of requests that they are seeing owning to this explosion in digital demand. A natural outcome of this digital trend is the emergence of AI-based virtual assistants, who can help professionals handle this load and allow them to focus on the most pressing problems. For example, in the case of medical professionals, they can focus on complex treatments and patient care, while the virtual assistants can manage the workload of answering frequently-asked questions.
Would it be easy to collate data across diverse sources, platforms and queries? How will AI solve the scale and precision parts of data? Will it also ensure that redundancy is addressed and real-time visibility into progress made in other areas is enabled?
The basic foundation of AI is data or information. Therefore, for AI to be successful, it is critical to develop an Information Architecture (IA). Such an IA can cater to data across diverse platforms, modalities, volumes, and quality. It also encapsulates important aspects about data governance and security. Such an IA then enables you to exploit the data to extract value out of it by developing AI models. Operationalizing these AI models also requires you to go through the process of model development, deployment, as well as monitoring these model decisions for various aspects such as accuracy, safety, and biases. Therefore, this journey from data to AI-based decision making is a step-wise ladder to be pursued.
Do we have any precedents or templates for similar problem-areas? Any special features that Watson would embrace for this specific task?
There are several examples where IBM's AI-based virtual assistants have helped professionals in their workloads. Most recently, at the Children's Healthcare of Atlanta in the state of Georgia, USA, Watson based ‘COVID-19 Pediatric Assessment Tool’ walks parents through a series of questions and results in suggested next steps that a parent should take. Here, Watson leverages recommendations on next steps according to the healthcare system's established protocols. Across the world about 25 countries are using the capabilities of Watson Assistant.
Can you share something on concerns around data privacy, security, lab confidentiality, recency, relevance etc?
As mentioned before, trust and security are at the very core of the journey to AI. Both at the IA level in terms of how data is captured, stored, cleansed; as well as on how the AI models are being used to make decisions and their robustness to adversarial attacks.
Is the main outcome here speed or collaboration? How will this assistant integrate with human factors in the process?
The virtual assistant is engaging constantly with human experts in all aspects ranging from gathering answers about medical best practises, to validating the quality of responses, to handing over complex situations to experts to manage.
What else is being envisaged at ICMR in terms of technology initiatives? Has technology been a help strong enough during this crisis—where exactly? Is there anything that could have been done better if we had that level or scale of technology ready in 2020?
We will be unable to comment on ICMR. However, at IBM—there are a range of scientific problems that are being addressed. For example, researchers are studying epidemiological models and cross-referencing them with weather patterns from weather data to understand and predict the underlying epidemiological spread better. Researchers are also looking at leveraging supercomputers to accelerate the speed of drug discovery and development. So, therefore technology has a very crucial role to play both during and after this pandemic is over.
Can you share some specific examples?
Some of the initiatives from IBM include a partnership with the US Department of Energy, IBM is leading the Covid19 High Performance Computing consortium to make a vast amount of supercomputing power (more than 400 petaflops of computing power, from ‘Peta’ means quadrillion and ‘flops’—floating point operations per second) available (like IBM Summit) to help researchers everywhere better understand COVID-19, its treatments and potential cures. The consortium recently selected to run an experiment from NIT Warangal on its supercomputers.
Another example is where IBM has pledged its entire global patent portfolio, which is comprised of more than 80,000 patents and patent applications, specifically, patent assets we feel are most relevant in the fight against COVID-19.