AI speaks letters, text-to-speech or TTS, text-to-speech, speech synthesis applications, creative … [+]
As the world's population grows and ages, health care systems in various geographies are teetering on the brink of collapse. According to the World Health Organization, the current number of health workers, including physicians, radiologists and other professionals, is insufficient to handle the growing caseload. On top of that, the stress and burnout caused by the increase in cases is forcing many people out of the field, further reducing the number of practicing workers. Baker Health estimates that about 72,000 U.S. doctors will leave the workforce between 2021 and 2022, and the roughly 30,000 who will join the workforce will not be enough to meet the growing demand.
Fundamentally, both of these challenges – increasing caseloads and shrinking workforces – are having one major impact: declining quality of patient care. This is where generative AI is much talked about, saving healthcare staff valuable time and resources and enabling them to focus on enhancing clinical outcomes.
Realizing the potential of generative AI
First, it's important to understand that AI is not new to healthcare. Organizations have been experimenting with predictive and computer vision algorithms for some time, particularly to predict the success of treatments and diagnose dangerous diseases before humans can. However, when it comes to creative AI, things are still pretty fresh, as the technology only came out a few years ago with the launch of ChatGPT. Gen AI models use neural networks to identify patterns and structures in existing data and generate new content such as text and images. They apply to all sectors, including healthcare – where organizations collectively generate around 300 petabytes of data every single day.
Now, with the ability to learn from data and create something new, gen AI may not completely replace doctors or do what they do, but it can certainly augment certain aspects of the system. Can simplify a strained health care pipeline. This can be anything from facilitating the patient journey and teleconsultation to managing medical documents and providing relevant information during the doctor's surgery.
Let's look at some of the most viable applications with gen AI that are possible today.
AI Assistant for Medical Guidance
After COVID-19, most organizations started teleconsultation services, where patients can contact a doctor without visiting a hospital in person. This approach worked but doctors were overworked as they had to deal with both online and offline patients. With gen AI, healthcare organizations can launch LLM-backed AI assistants to tackle this. Basically, they can fine-tune models like GPT-4 on medical data and build assistants that can take basic clinical cases and guide patients to the best treatment based on their system. If a particular case seems more complicated, the model can refer the patient to a doctor or a nearby healthcare professional. In this way, all cases will be resolved without putting the doctors under heavy work pressure. Several organizations, including Sanofi, Bayer, and Novartis, have taken this approach and launched AI assistants on their respective platforms.
Agents to streamline administrative work
Along with assessing situations and providing guidance, generative AI chatbots can also be built to handle basic healthcare tasks such as booking appointments and reminding patients about their scheduled visits. This can save hours of assigning human operators to handle the ever-increasing number of calls and messages in the healthcare system.
Providers using conversational AI agents include Mercy Health, Baptist Health, and Intermountain Healthcare. All of them have launched bots to automate tasks such as patient registration, routing, scheduling, FAQs, IT help desk ticketing, and prescription refills. In addition, many have also begun deploying general AI copilots that listen to patient-physician conversations and generate summary clinical notes, allowing physicians to manually enter information into the EHR. The hassle of documenting and filing is avoided. Nabla, one of the providers of such copilots, even uses these notes to prepare a set of patient instructions, on behalf of the doctor. This capability can be further developed into a general AI system that sits with a doctor and provides personalized treatments and therapies based on current conditions as well as previously recorded parameters including genetic make-up, health history and lifestyle. Makes plans.
Data retrieval in workflows
A major strength of LLMs is that they can be extended with retrieval augmentation generation (RAG) to tap additional data resources without retraining. This enables healthcare organizations to develop internal smart assistants or search systems that can provide highly relevant, contextual answers to any question. For example, RAG-based systems can help decision-support clinicians by providing evidence-based recommendations for a specific condition.
In other cases, they may generate factual clinical reports/patient data from the EHR system or share updated clinical guidelines for treatment. San Diego-based Nanom used this technique to develop an assistant that taps large language models (LLMs) and accesses real-time internal data and molecular simulation systems to help pharma teams design their drugs. Can help with development workflow.
Data analysis, report preparation
Another notable application of generative AI will be data analysis, particularly the analysis of medical images such as CT scans, MRIs and X-rays. Even with rapid digitization, most diagnostic agencies today rely on human experts to study medical images and write reports for patients. The work requires a lot of time and effort and is even prone to errors arising from inherent biases or just basic human fatigue.
With a general AI-powered approach, teams can optimize models like GPT-4 Vision and use them to study medical data and generate reports, automating and speeding up the entire process. Yes, the idea is still fresh, but early experiments show that this is a promising application of gen AI in healthcare. In fact, a JAMA Network study found that the quality and accuracy of AI-generated reports for chest radiographs are the same as those generated by human radiologists.
Drug development
Finally, with the ability to understand complex patterns and structures in complex medical data, creative AI can also aid in drug development. This technology can assess unique markers of a particular disease and come up with new combinations of chemicals or new molecular structures that could lead to potential drug candidates. It can even screen developed compounds based on their properties and predict side effects and drug interactions.
Just last year, Insilico Medicine's AI-generated drug INS018_055 for idiopathic pulmonary fibrosis, which affects about 100,000 people in the U.S., entered clinical human trials and is now in wide release.
Caution is required.
Despite these potentially transformative applications, healthcare organizations must realize that creative AI will only be as good as it's trained if the data isn't well-crafted or a If the type has a bias, then the results of the models will also reflect these problems, which will affect the reputation of the business. This means organizations must first prepare the data in the best possible way – while removing personally identifiable information (PII) from it – and then move on to the next stages of the project lifecycle, including training. and to minimize estimation, clinician and administrative effort. less burdensome.