AI In Healthcare - Challenges And Future

"Digital transformation since the pandemic has been massive. Telehealth has gone from being a novelty to a necessity. Having said that, we need to be reliant on healthcare institutions to get cured and we need technology to make it better," said Mitali Dutta, head of data science and predictive analysis, group IT information and data management, Philips Innovation Campus, at her session at The Rising 2021 by Analytics India Magazine.

She pointed at a rise in chronic diseases and healthcare costs, a scarcity of healthcare professionals in India. According to Dutta, the healthcare industry has to focus on the following four aspects, which she called Quadruple AIM of India’s healthcare system: 

• High efficient ecosystem to drive efficiency
• Personalised attention and better outcome
• Reduced cost of cure
• Reduced burden on healthcare staff

To achieve all the above, India needs artificial intelligence.

Role of AI in Medical

“AI can make medical care more human. Sometimes people believe AI replaces everything. I believe AI will in fact enable some of these clinical experts to find more time to spend with patients,” she said.

AI in Radiology 

The radiology departments aim to minimise exposure to the patients and get the scans right in the first try.

“For this, there are AI-related algorithms that get initiated by healthcare companies along with equipment that ensure high-quality images in a shorter span,” said Dutta.

Secondly, these equipment are too sensitive and if a patient is moving a lot, the scanner can’t take the image properly. So the staff need to control the motion. “To do that, we need to think of  software which can work even without making patients wear or use some physical equipment such as belts,” she said. 

Additionally, the biggest challenge today for radiology is making patients run from one department to another collecting reports. This also can be addressed by using AI-enabled workflow solutions.

Additionally, chest X-rays, accurate precision diagnostics, and providing the right measurements for ultrasound are some standard areas where AI is being leveraged. 

Challenges of AI in Healthcare

According to Dutta, the four challenges faced by the healthcare AI industry are:

1. The huge pressure on healthcare systems and equipment

2. Exponential growth of healthcare data

3. Producing perfect insights at the point of decision making

4. Augmented intelligence for the clinicians

5. Integration and legal challenges

“Interoperability and integration is one of the key factors that separates academic research from practical applications of AI and that’s what all the hospitals are looking at right now. They are setting up departments to deal with companies like us,” she said.

She also spoke about the need to connect the dots along the continuum of care. For instance, linking hospitals to home, primary care etc. She also believes medical equipment vendors have to start enabling the creation of applications by third parties like innovative startups or academic clinical centres through publishing application programming interfaces (API).

“We need APIs to consume data from the systems and that’s also a bit of a challenge for the industry right now,” she said. 

Lastly, she spoke about the legal challenges. “Needless to say, safeguarding patient privacy is very important. And without doing that the FDA doesn’t allow us to release any AI algorithms.” She went on to explain how the FDA regulations treat AI algorithms,  using the graph below. 

Future of AI in Healthcare

Elaborating on the future potential of AI in healthcare, she said, we must first acknowledge that we don’t have all the answers yet.

“In AI in healthcare, the centre is patient. So we also have to be careful about how transparent we are in addressing some of these AI solutions back at the hospitals and how much doctors are able to understand that,” she said. 

Having data systems to control and manage all the data, having effective data modules and data science techniques are also certain aspects that are quite essential, she said. “Deep learning lessons are also very important to make some of the images and illustrations readable and understandable for all data scientists,” she said.