Beyond The Hype: Three Best Practices When Using Generative AI In Healthcare
There's no denying that technology amplifies the power of humanity. We see it in our homes with automatically adjusting thermostats, in our cars with AI-powered brakes and in our phones with supercomputing capabilities. That also extends to the world of healthcare. The industry has invested deeply in tools to diagnose and treat medical conditions more accurately, which has helped to more than double life expectancy over the past century.
Now we need to adopt a new mindset around harmonizing how humans and machines work together. From my perspective, the largest challenge we have today globally in healthcare is patient access because we don't have enough physicians or nurses to serve everybody how we need to.
Enter generative AI. As reported by Becker's Hospital Review, research by Accenture, where I'm a global health industry lead, shows that 98% of healthcare providers and 89% of healthcare payerRead More : Could security be the prescription the healthcare industry needs? executives believe generative AI advancements "are ushering in a new era of enterprise intelligence." In today's workforce-challenged environment, generative AI's ability to liberate clinical resources from administrative burdens can allow healthcare professionals to focus on their core expertise and spend more time with patients and care teams.
Language-based AI has the potential to significantly shift tasks to support or augment approximately 40% of working hours, Accenture research also found. That's because generative AI is good at text classification, translation, summarization and question-answering. Large language models can learn and comprehend contextual patterns and relationships between words and sentences, producing creative responses, comprehending language structure and even suggesting intent.
In healthcare, generative AI can also streamline patient services, automate manual tasks and improve patient education. Clinicians spend a lot of time manually documenting information. Generative AI can take on the process of documenting care summaries, treatment plans and referrals. The technology can also simplify complex medical language into summaries that can be easily understood by patients and translated into multiple languages.
The big question today is how much of this is hype versus reality. I believe generative AI can be implemented today for the uses outlined above, but it will require several actions to get there.
1. Prepare the digital foundation.
The first priority is to get the digital foundation ready, including foundational systems and cloud-based data. Models for AI need data to learn, and that data must be accurate. Organizations need to take a strategic approach to ensure confidence in the underlying data. That means having clarity on where the data comes from, as well as having controls in place to safeguard it and deploy it in the right way.
One of the most important factors when it comes to applying generative AI in healthcare is ensuring that patient data always remains secure and confidential. Patient records contain highly sensitive information, so any systems using generative AI must be built with privacy and security measures in place to protect the data from unauthorized access or manipulation.
2. Establish controls for AI.
This is a priority at the outset of generative AI consideration, and a good way to start is with experimentation. Both clinical activities and internal administrative uses of AI could profoundly impact healthcare; however, patient-facing use cases come with a significantly higher risk where 99% accurate may not be good enough. By experimenting on a smaller scale first, organizations can learn from smaller use cases and deploy the right controls to ensure they're delivering responsible AI and avoid issues before scaling more broadly. Before deploying any generative AI system within a clinical environment, the solution should be subject to rigorous testing and validation processes to make sure it meets regulatory requirements for safety and efficacy.
3. Harmonize technology with the people it impacts.
Humans need to be the top priority because AI applications depend on people to guide them. AI can take over specific tasks, not entire jobs, becoming something of a co-pilot for humans. Whether using generative AI in a call center to respond to patient inquiries or during a doctor visit to assist with documentation, organizations need to reimagine the nature of work, focusing on activities that generate the most return on investment for people and the organization. It's not a matter of the capability of the technology itself; instead, I believe organizations should focus on how they will harmonize the technology with the work of humans to improve the experience of the clinicians and ultimately deliver better outcomes for patients.
There is a lot of optimism about the opportunity for generative AI to transform how healthcare is delivered—and for good reason. As this technology progresses and models improve, organizations need to focus on evolving operations and training people, as much as on technology investments. Helping people keep up with technology-driven change will be the biggest factor in creating greater access, ensuring better patient and clinician experiences and improving health outcomes for all.
Now we need to adopt a new mindset around harmonizing how humans and machines work together. From my perspective, the largest challenge we have today globally in healthcare is patient access because we don't have enough physicians or nurses to serve everybody how we need to.
Enter generative AI. As reported by Becker's Hospital Review, research by Accenture, where I'm a global health industry lead, shows that 98% of healthcare providers and 89% of healthcare payer
Language-based AI has the potential to significantly shift tasks to support or augment approximately 40% of working hours, Accenture research also found. That's because generative AI is good at text classification, translation, summarization and question-answering. Large language models can learn and comprehend contextual patterns and relationships between words and sentences, producing creative responses, comprehending language structure and even suggesting intent.
In healthcare, generative AI can also streamline patient services, automate manual tasks and improve patient education. Clinicians spend a lot of time manually documenting information. Generative AI can take on the process of documenting care summaries, treatment plans and referrals. The technology can also simplify complex medical language into summaries that can be easily understood by patients and translated into multiple languages.
The big question today is how much of this is hype versus reality. I believe generative AI can be implemented today for the uses outlined above, but it will require several actions to get there.
1. Prepare the digital foundation.
The first priority is to get the digital foundation ready, including foundational systems and cloud-based data. Models for AI need data to learn, and that data must be accurate. Organizations need to take a strategic approach to ensure confidence in the underlying data. That means having clarity on where the data comes from, as well as having controls in place to safeguard it and deploy it in the right way.
One of the most important factors when it comes to applying generative AI in healthcare is ensuring that patient data always remains secure and confidential. Patient records contain highly sensitive information, so any systems using generative AI must be built with privacy and security measures in place to protect the data from unauthorized access or manipulation.
2. Establish controls for AI.
This is a priority at the outset of generative AI consideration, and a good way to start is with experimentation. Both clinical activities and internal administrative uses of AI could profoundly impact healthcare; however, patient-facing use cases come with a significantly higher risk where 99% accurate may not be good enough. By experimenting on a smaller scale first, organizations can learn from smaller use cases and deploy the right controls to ensure they're delivering responsible AI and avoid issues before scaling more broadly. Before deploying any generative AI system within a clinical environment, the solution should be subject to rigorous testing and validation processes to make sure it meets regulatory requirements for safety and efficacy.
3. Harmonize technology with the people it impacts.
Humans need to be the top priority because AI applications depend on people to guide them. AI can take over specific tasks, not entire jobs, becoming something of a co-pilot for humans. Whether using generative AI in a call center to respond to patient inquiries or during a doctor visit to assist with documentation, organizations need to reimagine the nature of work, focusing on activities that generate the most return on investment for people and the organization. It's not a matter of the capability of the technology itself; instead, I believe organizations should focus on how they will harmonize the technology with the work of humans to improve the experience of the clinicians and ultimately deliver better outcomes for patients.
There is a lot of optimism about the opportunity for generative AI to transform how healthcare is delivered—and for good reason. As this technology progresses and models improve, organizations need to focus on evolving operations and training people, as much as on technology investments. Helping people keep up with technology-driven change will be the biggest factor in creating greater access, ensuring better patient and clinician experiences and improving health outcomes for all.
Source: www.forbes.com
