Revolutionizing Healthcare with Generative AI: The Creative Code for Staying Well and Spry
ChatGPT, an artificial intelligence, or AI, chatbot developed by OpenAI, has now launched its fourth version of what will probably be a long list of evolving iterations of a product that could disrupt almost every known industry. However, ChatGPT is just one of the many tools developed using generative AI. While iterations of generative AI technology have existed for decades, applications such as Midjourney and DALL-E (which create images from text) are milestones, especially in unsupervised machine learning (ML) and deep learning applications. Further developments in the quality and diversity of content and new generative model schemes will lead to broader adoption across different industries such as healthcare, insurance, finance, and more.
RELATED: Transforming Public Health Solutions via AI Revolutions
But what is generative AI? Generative AI describes algorithms capable of summarizing and creating new and unique data, images, audio, code, and even videos. Recent breakthroughs in the field can improve patient outcomes and enhance the efficiency of healthcare delivery. Until now, we have seen AI successfully applied to pattern recognition. This has been of great use for radiology, pathology, dermatology, and other fields of medicine where patterns are central to diagnosis.
So what can we expect from this new technology? Generative AI has several applications in healthcare, including drug discovery, medical imaging, disease diagnosis, and personalized medicine. In this article, we will explore the impact of generative AI in healthcare and how this technology can be used to improve patient outcomes.
Generative AI in drug discovery
Designing novel drugs is known to be challenging, expensive, and time-intensive. It takes approximately 12 to 14 years and costs around $3bn for a new drug to be brought to market. A significant portion of this overall cost and time is attributed to the drug discovery phase, which involves synthesizing thousands of molecules to identify a single pre-clinical lead candidate. Generative AI can speed up this process by analyzing vast amounts of data to identify patterns and predict outcomes to recognize potential drug targets and predict the efficacy of new drugs. By understanding biology and chemistry literature, new supercomputing-scale large language models (LLMs) empower scientists to gain unique insights into proteins, small molecules, DNA, and biomedical text. These large, powerful models learn from unlabeled data, such as sequencing data, on multi-GPU, multi-node, high-performance computing infrastructures. Companies like Nvidia, IBM, or Google’s DeepMind research project have taken strategic positions to conquer this space and disrupt drug discovery processes.
Benefits of AI in healthcare for disease diagnosis
Generative AI can be used to improve the accuracy and speed of disease diagnosis by analyzing patient data and creating new synthetic data that can be used to identify disease patterns. An excellent example of this would be AI-generated high-resolution images that could be used to train machine learning models meant to recognize patterns and diagnose. The same concept of building synthetic datasets applies to patient records, audio files, and more. This technology could represent a degree of degradation for data moats since the training of AI and ML models no longer depends on the data the developing team has access to. However, this doesn’t mean that access to high-quality, non-synthetic patient data will no longer be a significant competitive advantage for health diagnosis startups.
Moreover, it has the capacity to disrupt the way we access knowledge. A specialized generative AI tool has the potential to become the one-stop-shop physicians will use to efficiently seek answers to questions they would otherwise need to search Google or other curated knowledge websites.
Its remarkable fluency and competent prose can effectively convey any idea to any audience, including patients, insurance companies, or clinicians. Many of us will have read about Chat GPT passing official examinations, including the United States Medical Licensing Exam or patients using the OpenAI product for self-diagnosis. Despite the potential for development in this field, the emergence of unscrupulous actors capitalizing on hype and confusion to promote unsafe or even unethical products is a concern — case in point: Martin Shkreli's medical ChatGPT knock-off, Dr. Gupta. While the name raises an eyebrow, the real issue lies in the danger of untested products entering the market.
Creating personalized medicine via generative AI
Personalized medicine can help to improve patient outcomes and reduce healthcare costs by providing more targeted and effective treatments. Generative AI can create personalized medicine by analyzing patient data and developing treatment plans or recommendations based on each patient's unique characteristics. This technology can also predict the outcomes of different treatments, thus reducing the need for live trial and error. This requires vast amounts of data collection and analysis, including medical records, lab tests, imaging tests, omics data, wearable devices, and patient-reported data. This data and the insights produced from it need to be communicated to the patient and the doctor. Generative AI can help with these processes by generating text or speech that can summarize information, explain diagnosis or therapy options, provide recommendations to healthcare professionals, or ask relevant questions to patients.
Download: The Future of Digital Health
One key aspect of personalized medicine is the use of real-world data. One of the most common ways of collecting this data is through wearable technology. The data collected through these devices can not only be recorded 24/7. Still, they could also be leveraged in real-time and complement other data sources, such as medical records, diagnostic tests, etc.
The use of AI to achieve personalized medicine has been introduced previously. It has been used for over two decades, and the progress and use cases are vast. Lately, we have seen many uses of AI attributed to generative AI. In reality, non-generative AI is capable and more fit for specific tasks than generative AI. A common example of this is AI algorithms used for pattern recognition in diagnostic imaging, which is able to increase accuracy and significantly reduce the time for diagnosis dramatically. Generative AI, however, can help overcome a few challenges in personalized medicine, such as different reliability and accuracy standards from wearables, by generating synthetic or augmented data that can complete the data gaps or existing data.
Automating clinical functions — AI in healthcare examples
Automating clinical functions will allow doctors to refocus on patient care. In the past 50 years, how clinicians see patients has changed dramatically. In the ’70s, a routine visit to see a doctor was typically scheduled for 30 minutes, and a new patient visit was usually booked for 1 hour. As access to healthcare expanded and the industry developed, the human side of care began to shrink. Fortunately, there are ways in which new technologies, such as generative AI, can help automate certain job functions and free up providers’ time, which could allow for better work-life balance and more extended visits with patients. These two points are crucial to improving healthcare. As of late 2022, 30% of doctors in the US say that they feel burned out, and a staggering proportion have also considered leaving their clinical care profession within the past six months.
In contrast, a 2018 report by Elena Andreyeva and fellow researchers at the University of Pennsylvania, published in the National Bureau of Economic Research, examined the impact of home health visit duration on patients recently discharged from hospitals after acute treatment. The results of this study pointed out that for every extra minute that visit lasts, there was a reduction in risk admission of 8%. For part-time providers, this number doubled to a 16% reduction, and for nurses, it was a 13% reduction per extra minute. A natural use for generative AI in the clinical setting is to reduce the amount of paperwork and streamline administrative tasks that contribute to this burnout. In this sense, we have seen Microsoft make a significant bet on OpenAI as a continuation of their healthcare strategy after their first step to address health data interoperability with Azure API for FHIR in 2019 and Microsoft Cloud for Healthcare, launched in September 2020 to address the health industry’s challenges — from reducing clinician burnout, delivering more personalized experiences for patients, and enabling health data interoperability.
Challenges in using generative AI
Data privacy and security concerns regarding challenges to implementing generative AI are at the top. Healthcare data is highly sensitive, and there is a risk that this data could be misused or accessed by unauthorized parties.
In the last couple of months, the Federal Trade Commission has fined two digital health companies under the premise of sharing the personal health information of millions of customers with third-party advertisers. Meanwhile, in Europe, GDPR is already starting to raise flags for chatbot/large language model (LLM) developers such as OpenAI, which has already been shut down in Italy by the country’s data protection authority and provided the company with a long checklist to complete for the suspension to be lifted. A relevant reminder to include at his point is that GDPR violation penalties can account for up to 4% of a company's global annual turnover or €20 million, whichever is higher.
When it comes to keeping data well-kept and secured, hospitals in the US are currently struggling with cyber attacks that breach the security of their systems. Out of the 14 most significant healthcare data breaches in US history, six happened last year, in 2022. To address this issue, healthcare providers must implement robust data security measures and ensure authorized parties only access patient data.
As mentioned before in this article, LLMs could be used by bad actors, perpetuate biases, or just point blank hallucinate. Since this technology follows a black box model, how can we trust the answer without knowing its reasoning process? For some areas of business or life, "magic that works" might be good enough, but that's not the case for health. If a decision is going to affect anyone's health, it should be crystal clear why and how it is being made, and this still requires a health expert to be centrally involved.
A related challenge is the need for skilled professionals to develop and implement generative AI technologies. Healthcare providers must invest in training and development to ensure their staff have the skills and knowledge to use generative AI effectively.
Healthcare is transforming through generative AI
Generative AI has the potential to transform healthcare by providing industry leaders with new tools and insights that can be used to discover new medicines and bring them to market faster, improve patient outcomes, and enhance the efficiency of healthcare delivery. By analyzing patient data and creating new data that can be used to predict outcomes and identify the best treatment options, the healthcare industry can provide more targeted and effective treatments, reducing healthcare costs.
NEXT: Top 10 Digital Health Startups in 2023
Pharmaceutical and biotech companies will be able to accelerate drug discovery and deliver personalized and more efficient treatments. At the same time, easier access to the latest data facilitates medical researchers staying up to date on the latest findings. Payers can benefit from analyzing past claims and creating new data matching identified patterns, enriching sales experience, or better customer service. While several challenges need to be addressed, mainly related to cybersecurity issues and data privacy, the opportunities presented by generative AI are significant. Healthcare organizations that invest in this technology will be well-positioned to succeed in the future.
Are you a corporate partner looking to connect with new healthcare technologies? At Plug and Play, we fast-trackinnovation in the healthcare industry. Reach out to connect with the most disruptive startups.