Just as quickly as they hit healthcare, generative artificial intelligence and large language models are reshaping the healthcare landscape. And CIOs and other healthcare IT leaders at hospitals and healthcare systems must fully understand these technologies before deploying them.
One real-world application of AI that is critical for healthcare organizations to understand: the use of AI-powered language models in communications between physicians and patients.
These models have been found to provide valid responses that simulate empathetic conversations for patients, making it easier to manage difficult interactions. But there are still many challenges that need to be overcome before the many applications of AI can move forward.
For example, one challenge is ensuring regulatory compliance, patient safety and clinical efficacy when using AI tools.
Dr. Bala Hota is senior vice president and CIO at Tendo, a healthcare software company focused on artificial intelligence. We interviewed him to discuss the concept of generative AI and large language models, deploying LLMs for healthcare applications, real-world applications of genAI, and challenges and ethical issues.
Q. CIOs and other IT leaders in hospitals and healthcare systems need to understand generative AI before deploying it. What are some things about genAI that you think are most important for these leaders to understand?
A. It is important for CIOs and IT leaders to understand that genAI is just one aspect of the broader digital transformation needed across the industry, and it is essential to understand the fundamental evolution that AI has undergone in recent years.
Generating data, amplifying and detecting anomalies can significantly accelerate decision-making within an organization. However, generative AI cannot replace human judgment and interaction. Instead, it acts as a supplement that can increase productivity.
The semantic component of large language models dramatically reduce the time an organization’s teams spend cleaning and presenting data, allowing them to operate optimally and focus on strategic tasks. Any form of AI must ensure adequate security, compliance and a common sense approach to data protection and distribution. The industry must ensure that the technology exceeds its practical applications.
Q. How can hospitals and healthcare systems best leverage large language models today?
A. The use of AI is becoming increasingly important in healthcare as it can help hospitals and healthcare systems streamline their decision-making processes, increase efficiency and improve patient outcomes. AI has a wide range of applications, from simplifying data to interacting with patients, which could have a significant impact on the healthcare industry.
A key benefit of AI in healthcare is improving the effectiveness of treatment planning. Ambient voice can be used improve the use of electronic health records. Currently, AI writers are being implemented to assist with medical documentation. This allows doctors to focus on patients while AI takes care of the documentation process, improving efficiency and accuracy.
Additionally, hospitals and healthcare systems can use AI’s predictive modeling capabilities to risk stratify patients, identify patients at high risk or increasing risk, and determine the best course of action.
In fact, AI’s cluster detection capabilities are increasingly used in research and clinical care to identify patients with similar characteristics and determine the typical course of clinical action for them. This could also enable virtual or simulated clinical trials to determine the most effective treatment courses and measure their efficacy.
Q. What are some practical applications of AI that you think will lead the way for the rest of the industry?
A. One real application of AI that is leading the way is the use of AI-powered language models in doctor-patient communication. These models have been found to provide valid responses that simulate empathetic conversations for patients, making it easier to manage difficult interactions.
This application of AI can significantly improve patient care by enabling faster and more efficient triage of patient messages based on the severity of their condition and message.
Furthermore, AI can be used for better risk stratification at the time of treatment. This can help healthcare providers deliver top performance by making better use of resources. By accurately identifying patients who require more intensive care, healthcare providers can deploy their resources more effectively and improve overall patient outcomes.
This includes automating patient interactions to scale communications and increase patient engagement. AI is used to reach patients with reminders, follow-ups and better engagement, leading to better outcomes. By identifying patients who require more intensive care, AI can help overcome barriers such as clinical inertia and poor treatment compliance, significantly improving outcomes.
Q. What do you think are the challenges and ethical issues of AI that healthcare providers need to address?
A. A challenge in implementing AI in healthcare is ensuring regulatory compliance, patient safety, and clinical efficacy when using AI tools. While clinical trials are the standard for new treatments, there is debate about whether AI tools should follow the same approach. Some argue that mandatory FDA approval of algorithms is necessary to ensure patient protection.
Another concern is the risk of data breaches and compromising patient privacy. Large language models trained on protected data can potentially leak source data, posing a significant threat to patient privacy. Healthcare organizations must find ways to protect patient data and prevent breaches to maintain trust and confidentiality.
Bias in training data is also a critical challenge that needs to be addressed. To avoid biased models, better methods must be introduced to avoid biases in training data. It is critical to develop training and academic approaches that enable better model training and integrate equity into all aspects of healthcare to prevent bias.
To address these challenges and ethical concerns, healthcare provider organizations must focus on developing datasets that accurately model healthcare data while ensuring anonymity and de-identification.
They should also explore approaches for decentralized data, models, and trials that leverage federated, large-scale data while protecting privacy. Additionally, partnerships between healthcare providers, healthcare systems, and technology companies must be established to put AI tools into practice in a safe and thoughtful manner.
By addressing these challenges, healthcare organizations can unleash the potential of AI while maintaining patient safety, privacy and fairness.
Follow Bill’s HIT coverage on LinkedIn: Bill Siwicki
Email him: bsiwicki@himss.org
Healthcare IT News is a HIMSS Media publication.