Because of significant increases in life expectancy, the world's population is aging at a startling rate. The number of people in the world who are 60 years of age or older will rise from 1 billion in 2020 to 1.4 billion in 2030, making up 16.7% of the total population. In 2050, this number is expected to have doubled to 2.1 billion, according to WHO. According to reports, 81.5 per cent of those over 85 years have at least two chronic diseases, and 92% of older persons have at least one chronic disease.  

The primary cause of multimorbidity and chronic diseases is the acceleration of ageing. As older individuals age, their unmet healthcare demands will inevitably rise, adding to the already heavy load on the healthcare system. Finding long-term methods to encourage care in this age range is, therefore, essential. 

Because of its ability to harness the power of extensive data, obtain insights to support evidence-based clinical decision-making, and enable value-based care, Artificial Intelligence is flourishing in the healthcare industry. AI adoption promotes health policy and planning, disease diagnosis and treatment, risk assessment for morbidity or mortality, and disease prediction and surveillance. 

Benefits of generative AI in healthcare 

The field of deep learning, and large language models specifically, has great potential to revolutionize the way clinicians provide care to the elderly population. The capacity of LLM programs, like ChatGPT, to produce human-like responses in response to a conversational cue opens up new avenues for engaging with and deriving insights from data, simplifies daily activities and automates mundane labour for physicians.  

Previous research has found how well LLMs perform time-consuming and laborious jobs involving little clinical decision-making, like handling tasks and messages in the electronic health record (EHR) system's communication hub. 

A group of medical professionals was asked to review ChatGPT's responses to questions along with physicians' responses to the same questions. They consistently rated ChatGPT's responses as higher quality and more empathetic than those composed by the physicians, demonstrating to researchers at the University of California, San Diego, that ChatGPT could respond to patient messages effectively. Several other universities are testing a similar strategy, either on their own or in collaboration with EHR suppliers. 

Using these techniques to improve clinical thinking and decision-making may be one of the most exciting uses of generative AI. Massive volumes of unstructured data can be ingested and synthesized by LLMs. This implies that almost all data contained in electronic health records (EHRs)—such as lab findings, imaging scan results, genetic information, and patient-generated health data—may be analyzed by licensed lifelong learners (LLMs). For instance, a busy hospitalist may find it difficult to summarize every detail in a patient's chart during admission. 

Nonetheless, medical professionals can identify the most valuable aspect of a patient's narrative through practice. Combining this clinical expertise with LLM-based methods makes it possible to find patterns, correlations, and subtle links in the clinical data that might not be immediately obvious. Consequently, this method can assist medical professionals in working more productively and successfully, as well as in making more precise and data-driven diagnoses. To aid in the creation of individualized preventative treatment plans, LLMs can also be used to find trends linked to high-risk patients with chronic illnesses. 

Enhancing patient experience

Patient experience is yet another top innovation objective in healthcare. For older adults who require help with personal and medical care and frequently confront health issues, ChatGPT can be a great source of information and support. Chatbots and virtual assistants with generative AI capabilities can remotely monitor high-risk senior citizens with chronic illnesses and offer individualized guidance on fitness, diet, and health to help them manage their ailments. ChatGPT can also assist in addressing social isolation and loneliness in older adults by providing a sense of virtual companionship, connections, and nonjudgmental emotional support. Future effects and impacts are anticipated from the creative applications of generative AI to improve health care for older adults. These applications, which include personalized prevention of cognitive decline, mental health support, and remote health monitoring, have been investigated in the literature more and more. 

Risks and limitations 

No generative or predictive model can perform perfectly, even with these technologies' unique powers. Understanding the causes of bias and mistakes in AI tools is essential, as is creating practical performance criteria that ensure safety. For example, "general knowledge" makes up most of the training data for the most extensive existing LLMs; these models are trained using an enormous and diverse data set sourced from the internet. Because of this, these models perform exceptionally well in a wide range of activities, but they may not be sufficient when specific medical expertise is needed.  

These models can unnervingly malfunction in opaque or misleading ways, which raises the question of whether they can truly assist in clinical decision-making. Furthermore, there may be significant expenses associated with employing these models, such as the direct cost of gaining access to them through a vendor or other third-party platform or the costs associated with developing, implementing, or maintaining open-source goods internally. Lastly, because many LLMs require a lot of energy and resources to operate, there are serious worries about how widespread use of these tools may affect the environment. 

Source: NIH

Image: Unsplash