Lessons Learned: AI’s Potential and Challenges in Transforming Healthcare

Our Take:

The first AI models for medical use were approved by the FDA in 1995. While only 50 were approved over the first 18 years, an average of about 3 per year, over 200 were approved in the year 2023 alone. As noted by Timothy Bates, clinical professor of cybersecurity in the College of Innovation and Technology at the University of Michigan-Flint in Venture Beat, “The integration of AI into healthcare is not just an evolution but a revolution that holds the promise of significantly enhancing patient care, operational efficiency, and medical research.”

For example, AI has the potential to cut annual U.S. healthcare costs by $150 billion by 2026 as it enhances drug discovery, development, and medical research. In radiology, AI can improve care, reduce costs, and alleviate radiologists' workloads, addressing the shortage of radiologists while ensuring patient safety and regulatory compliance. AI's ability to outperform traditional CAD software by decreasing false-positive marks per image by 69% and effectively identify patients at high risk of postoperative adverse outcomes showcases its potential. Similarly, in cardiology, AI improves outcomes, increases accessibility, and enhances efficiency, with continuous monitoring and predictive analytics preventing cardiac emergencies while AI offers solutions to alleviate anesthesiologist shortages and enhance healthcare delivery efficiency.

Despite these benefits, the integration of AI into healthcare raises significant ethical and operational challenges. The potential for bias in AI algorithms can lead to disparities in healthcare delivery, where certain populations may receive less effective care based on biased data inputs. Privacy and security concerns are paramount, as healthcare data involves sensitive patient information that must be protected against breaches and unauthorized access. Regulatory compliance is also a major challenge, as healthcare providers must navigate a complex landscape of guidelines that govern the use of AI in clinical settings.

Key Takeaways:

The first AI models for medical use were approved by the FDA in 1995, with only 50 approved over the first 18 years, while almost 200 were approved in 2023 alone (Encord)
One study found that a widely used model of health risk reduced the number of Black patients identified for extra care by more than half due to racial bias (Science)
An automated machine learning model that analyzed a dataset of over 1M patient encounters effectively identified patients with high risk of postoperative adverse outcomes, illustrating the potential for AI to enhance risk prediction in surgical settings (Journal of the AMA)
AI applications have the potential to cut annual U.S. healthcare costs by $150 billion by 2026 as AI is used more for drug discovery and development, and improving medical research (Accenture)

The Backdrop:

As artificial intelligence (AI) continues to advance, its integration into various sectors is transforming traditional practices and pushing the boundaries of what is possible. One area where AI's impact is particularly profound is healthcare, where it is revolutionizing drug development, medical diagnostics, and patient care. By leveraging AI's capabilities, the healthcare industry stands to benefit from more efficient processes, reduced costs, and improved patient outcomes. However, the rapid adoption of AI also brings significant ethical and regulatory challenges that must be addressed to ensure its safe and equitable implementation.

Regulatory bodies play a crucial role in shaping the use of AI within healthcare. The U.S. Food and Drug Administration (FDA) has established guidelines to ensure that AI technologies are both safe and effective before they are marketed. The European Union’s General Data Protection Regulation (GDPR) mandates transparency in AI algorithms, requiring clear explanations of how patient data are used and decisions are made, which is essential for building trust in AI technologies.

Looking to the future, the potential for AI in healthcare is vast, with expectations for even greater technological integration. Continuous advancements in AI technology are likely to lead to more sophisticated healthcare solutions that are personalized and efficient. However, the success of these technologies will depend on their acceptability among healthcare providers and their ability to integrate seamlessly into existing healthcare systems. Future AI tools will need to be more user-friendly and aligned with the practical and clinical needs of healthcare workers to enhance their usability and effectiveness.

For healthcare administrators and policymakers, the strategic implementation of AI involves promoting education and training to enhance understanding and acceptance of AI technologies among healthcare providers. Additionally, developing ethical frameworks to address issues of bias, privacy, and security is crucial. International collaboration will also be essential to ensure that AI technologies can be harmonized across different regulatory environments, facilitating global access to the benefits of AI in healthcare.

Lessons Learned:

What have been some of the “lessons learned” around increasing the implementation and effectiveness of AI technology and access to health treatment and prevention from our prior “Our Take” posts over the years?

While AI holds great potential it brings with it widespread ethical concerns over the use of AI in healthcare around potential biases, data privacy risks, lack of transparency, and job displacement. This will necessitate ongoing collaboration and updated guidelines to ensure informed consent, fairness, and patient trust.

As of January 2023, there were 520 FDA-cleared AI algorithms, approximately 396 of which were for radiology and 58 of which were for cardiology (Radiology Business)
One study found that a widely used model of health risk reduced the number of Black patients identified for extra care by more than half due to racial bias (Science)
The first AI models for medical use were approved by the FDA in 1995, with only 50 approved over the first 18 years, while almost 200 were approved in 2023 alone (Encord)
AI applications have the potential to cut annual U.S. healthcare costs by $150 billion by 2026 as AI is used more for drug discovery and development, and improving medical research (Accenture)

Navigating the Ethical Landmines of AI in Healthcare

Making AI models more understandable and transparent, can be implemented with improved design and auditing to help identify and mitigate biases. This will help increase accountability and foster trust, which can accelerate AI adoption

AI algorithms continuously adjust the weight of inputs to improve prediction accuracy but that can make understanding how the model reaches its conclusions difficult. One way to address this problem is to design systems that explain how the algorithms reach their predictions.
ChatGPT4 is rumored to have around 1 trillion parameters compared to the 175 billion parameters in ChatGPT3 both of which are well in excess of what any human brain could process and break down.
During the Pandemic, the University of Michigan hospital had to deactivate its AI sepsis-alerting model when differences in demographic data for patients affected by the pandemic created discrepancies and a series of false alerts.
AI models used to supplement diagnostic practices have been effective in biosignal analyses and studies indicate physicians trust the results when they understand how the AI came to its conclusion.

Explainable AI-Making AI Understandable, Transparent, and Trustworthy

In radiology, AI has already been shown to reduce costs and alleviate radiologists' workloads by assisting in image analysis, identifying abnormalities, and automating tasks. While AI is positioned to address the growing shortage of radiologists its adoption must be managed carefully to ensure patient safety and it’s seamless incorporation into existing clinical workflows.

Between 2010 and 2020, the number of diagnostic radiology trainees entering the workforce increased by just 2.5% compared to a 34% increase in the number of adults over 65 (RSNA)
In one study, comparing AI-CAD and traditional CAD software, the AI system outperformed by decreasing the false-positive marks per image (FPPI) by a significant 69% (Diagnostics)
Over 85% of outpatient facilities and hospitals are facing staffing challenges, while they’re anticipating a 10% uptick in demand for staffing across MRI, nuclear medicine, ultrasound, radiologic and cardiovascular technologists (U.S. DOL)
In one study from the Netherlands of over 40K women with extremely dense breast tissue, scanning using commercially available AI software led to significantly fewer interval cancers than the control group (Pediatric Radiology)

AI in Radiology: Aiding Workflows and Accuracy as Workforce Pressures Mount

The integration of AI into drug development has the potential to significantly reduce costs, shorten development times, and increase the efficiency. This could dramatically increase the effectiveness of the development process, ultimately leading to better treatments and more personalized healthcare.

Most drugs take between 10-15 years to be developed at an average cost of $1-2B before receiving [U.S.] approval for clinical use (Chinese Academy of Medical Sciences and the Chinese Pharmaceutical Association)
It is estimated that 85% of the human proteome is considered undruggable and finding effective pharmaceuticals to target these proteins is considered exceptionally hard, or impossible (The Cambridge Crystallographic Data Centre)
Machine learning methods such as eToxPred correctly predict synthetic accessibility and toxicity of drug compounds with accuracy as high as 72% (BMC Pharmacology and Toxicology)
The use of Machine Learning in drug discovery could save approximately $300-400M per drug (U.S. General Accounting Office)

4 Ways AI Could Revolutionize The Future of Drug Development…No Chat Needed

AI improves outcomes, increases accessibility, and enhances efficiency in cardiology via the application of continuous monitoring and predictive analytics. Applying new AI-based treatments has the potential to reduce morbidity and create new treatment protocols.

6.2 million adult Americans have heart failure, with prevalence projected to increase by 46% and direct medical costs escalating to $53 billion by 2030 (CDC, Journal of Managed Care & Specialty Pharmacy)
Between 2017 and 2020, almost 128M US adults had some form of cardiovascular disease with total costs of $407.3B (American Heart Association)
Despite improvements in the treatment and incidence of heart failure the 1-year mortality rate remains approximately 30%, while the 5-year mortality rises to 40% (Circulation)
Cardiovascular disease was the underlying cause of death, accounting for almost 1M deaths in the United States in 2020 (American Heart Association)

Using AI in Cardiology to Soothe the Heart

As surgical demands increase with an aging population, AI offers solutions to alleviate anesthesiologist shortages and enhance healthcare delivery efficiency.

In 2020, workplace staffing shortages affected 35.1% of US anesthesiologists, a figure that rose to 78.4% in 2022 (Anesthesiology)
Experts project approximately a 100% increase in Americans aged 50 years and older with at least one chronic disease by 2050 leading to an increase in patient complexity and surgical demand (Frontiers in Public Health)
An automated machine learning model that analyzed a dataset of over 1M patient encounters effectively identified patients at high risk of postoperative adverse outcomes, showcasing the potential for AI to enhance risk prediction in surgical settings (Journal of the American Medical Association)
Anesthesiologists successfully used a closed-loop system to maintain blood pressure within 10% of a target range for >90% of the case time in patients undergoing abdominal surgery (Journal of Personalized Medicine)

AI in Anesthesiology: Lowering the Risk of Surgical Complications and Adverse Outcomes

Final Thoughts:

The future of AI in healthcare is promising yet complex. While its potential to revolutionize drug development and medical practices is immense, careful consideration must be given to ethical, regulatory, and practical challenges. Ensuring transparency, fairness, and patient trust will be crucial as AI technologies become more integrated into healthcare systems. By fostering collaboration among technologists, healthcare professionals, and policymakers, we can harness AI's full potential to create a more efficient, effective, and equitable healthcare landscape. As we move forward, it will be essential to balance innovation with responsibility, ensuring that the benefits of AI are realized for all patients.

Related Readings: