Sector-wide debates about artificial intelligence and the use of platforms like ChatGPT, with enthusiasts touting recent breakthroughs while AI’s potential pitfalls and technology could be exploited While pointing out the method, it’s overheating.
And while AI is already being deployed in some areas of the healthcare sector, there are significant hurdles facing the industry as it seeks to bring its latest technological advancements to the field.
Former Salesforce Chief Medical Officer Geeta Nayyar likened these challenges to people facing self-driving cars. Not everyone has their hands up for him to get into a car driven entirely by AI.
“The idea of a self-driving car that complements the actual person in the driver’s seat right now is interesting. That’s what healthcare needs,” Nayyar said in an interview with Yahoo Finance.
“Everyone is obsessed with this: ‘We don’t need doctors anymore… AI’s here to help.'” That’s not true, she added.
The topic was the focus of a major healthcare conference this month, where executives warned there is still much to understand about machine learning and its implications for the sector.
One of the biggest uptake challenges facing the healthcare sector is industry-specific. Most healthcare data is largely protected and hidden. This privacy also complicates efforts to build robust enough data pools to deploy across sectors.
The US government has already expressed concerns about the role of AI in healthcare and has determined that data quality is a key area that needs improvement.
According to a Government Accountability Office report, “developing or expanding access to high-quality datasets will help facilitate training and testing of ML (machine learning) techniques in diverse and representative conditions. There is a possibility”. “This will improve the performance and generalizability of our technologies, help developers understand performance and areas of improvement, and help build trust and adoption for these technologies.”
Today, more than 97% of health data entered into the system is unused. That data volume has grown from 15% of all health information digitized in just 10 years to almost 98% today. GE Healthcare (GE) Chief Medical Officer Taha Kass-Hout told Yahoo Finance.
However, in Kass-Hout’s view, this data lacks structure and is still unreliable. ”[A]Without it, we can’t efficiently query, analyze, and make decisions,” says Kass-Hout.
A “much more efficient” tool
Kass-Hout remains optimistic about AI’s use in health, noting that AI is already being used in radiology and creating new treatments, for example.
But the continued involvement of humans in these processes is key to AI’s success, he said. Nayyar agreed, saying AI could also be used as a tool to keep patients on their medication. She reiterated the example of self-driving cars.
“I don’t necessarily have to drive for you, but can I put you in the right car and tell you to put your seatbelt on?” Nayyar said. “There are many things AI can do to complement the patient’s journey.”
AI also has the potential to help make research and development more efficient in the pharmaceutical industry. For example, Moderna (MRNA) used her AWS (AMZN) to identify vaccine formulations to pursue in efforts to protect against COVID-19.
said Dave Johnson, vice president of information science, data, science and AI at Moderna.
And the biotech giant has announced a new partnership with IBM (IBM) to further explore the use of AI in medicine.
In addition to helping pharmaceutical companies, Bill Gates recently touted the potential of AI for the potential this technology could have for public health in developing countries.
“We already have software that can look at this data and infer what the pathway is, search for pathogen targets, and design drugs accordingly,” Microsoft (MSFT) founders said in a recent letter: says so. Gates says the next generation of his AI tools will be much more efficient and “will be able to predict side effects and understand dosage levels.”
Some experts hope AI can help with paperwork, too. This is a significant burden for clinicians in the era of electronic medical records and claims, and is also critical for insurance companies that determine claims coverage.
Bad data can ‘endanger progress’
Experts are also concerned about the potential for AI to generate results based on distorted data that machines learn. This is a known risk of machine learning models in medicine.
Addressing these issues would also help allay clinicians’ concerns, notes the GAO report. And a well-known route — robust clinical trials — could be the solution.
“Despite the excitement and proven benefits of introducing (AI) tools into clinical practice, poor data quality and common biases in healthcare jeopardize progress toward achieving health equity.” and may contribute to continued uncertainty and hesitation about adopting these tools,” the report said.
Recent discussions have focused on advancing technology too fast, as highlighted in a recent letter signed by Tesla (TSLA) CEO Elon Musk and Apple (AAPL) CEO Tim Cook. That’s why we are here. AI development in general doesn’t have to be put on hold, but at least for six months “regressing from dangerous competition to ever-larger, unpredictable black-box models with new capabilities.” “If such a moratorium is not enacted promptly, the government will have to step in and set a moratorium,” the letter said.
But others, like Gates, say calling for a pause is impractical. Instead, we need responsible buildings, Kass-Hout said. This is easy in the world of health, thanks to regulations already in place. That means figuring out how the model behaves over time, how to measure results, and “evaluating, modifying, editing in a loop,” Kashout added.
And the data should reach robustness, i.e. multimodal. This includes storing personal medical notes, texts, medical images, records, and test results all in one place to aid in analysis and diagnosis.
All of this, in Kass-Hout’s view, means that it will take time for the technology to make a significant impact in the medical field. “These models have never been trained on this kind of data.”
follow anjali Twitter @AnjKhem
Click here for the latest stock market news and in-depth analysis, including stock-moving events
Read the latest financial and business news from Yahoo Finance
