Understanding generative AI and ChatGPT in pharma

What is generative AI?

Generative AI is a form of machine learning that can product text, video, images, and other types of content in the same medium in which it is prompted, such as text to text, or in a different medium from the prompt, such as text to image.

Applications for generative AI include creative and academic writing, translation, sound editing and image editing. Experts have raised concerns about generative AI, citing legal, ethical, political, social, and economic issues.

AI use cases in pharma

Narrow AI systems can parse and analyze enormous data sets with far greater speed, accuracy, and efficiency than even the most highly-trained humans, making them powerful tools for enhancing the drug development process.

“[AI is] not a replace everything, replace the entire department with AI and technology [proposition]. Instead, consider how to optimize one part of the workflow in the life science industry [especially for medical affairs].” – Jason Smith, Chief Technology Officer, Within3

Here’s how pharma and AI can intersect.

Big data and life science

The term ‘big data’ describes data sets of such significant volume that they can’t be stored or processed by traditional data processing software. The field of big data exploded during the digital revolution of the early 20th century, and the life science sector produces more big data sets than almost any other. Everything from research data to electronic medical records (EMRs), clinical trial data, and the information captured by wearable technology is growing exponentially – providing both obstacles and opportunities for life science companies.

In a process known as hit identification, AI algorithms are used to search vast libraries of existing compounds to determine whether they could serve as active ingredients in new drugs. And in a similar process called precision medicine, AI programs churn through complex disease data sets, looking for patterns that might lead to potential treatment modalities. Genetic Engineering and Biotechnology News report, “The role of AI is to ‘connect the dots’ for enormous amounts of clinical, genomic, and patient data to find the potential utility of an existing drug for a certain condition or disease.”

AI systems are already widely used to help identify and generate novel drug candidates, repurpose existing drugs, and validate potential drug candidates. There’s also an opportunity for pharma AI and machine learning to crunch manufacturing process data and to make these processes more efficient – minimizing expense and reducing time to market.

Our AI-powered Moderator Assistant gets you more feedback with 90% less work. Learn more

Clinical trial optimization

The clinical trial process is long-winded, expensive, and often fails. But AI is already helping to make clinical trials more efficient, bring new treatments to market more quickly and maintain patient safety.

By parsing diverse EMR data, AI is improving trial recruitment by identifying populations with the best chance of responding to treatment. AI is being used to replace placebo control groups with ‘digital twins’ of subjects for randomized clinical trials, allowing clinical teams to reduce the size of their control groups and minimize patient disruption. AI applications are being used to clean, aggregate, store, and manage clinical big data – helping clinical teams inform better site selection, optimize study designs, and accelerate the informed consent process.

Predictive AI models can be used to calculate the results of a given treatment on a subject or group of subjects – potentially removing the need for animal testing and accelerating the preclinical phase. And finally, AI-powered wearable technologies are proving invaluable to clinicians as patient monitoring tools, enabling “automatic detection of physical and emotional states.” When asked what technologies would improve clinical trial efficiency most, “the top answer from survey responses was leveraging big data and AI.”

Diagnosis and disease identification

As mentioned, narrow AI applications are far more adept than humans at finding patterns, anomalies, and correlations in big data sets. This ability is beneficial for accurate diagnosis and disease identification. Soon, AI is expected to minimize human error in disease identification and make more accurate diagnoses faster and earlier than human HCPs – leading to significantly improved patient outcomes.

Most experts suggest that we shouldn’t view AI as something that’s going to replace doctors, but as a tool to support them – potentially creating a more equitable healthcare landscape capable of bringing quality care to more people around the world.

Medical imaging

Diagnostic imaging represents one of the most promising avenues for AI in life science. A recent study revealed that narrow AI is already on par with human HCPs in analyzing medical images. Given that the rate of AI growth is expected to be exponential, this has significant implications for specialisms such as radiology, dermatology, pathology, and ophthalmology. AI is so efficient at differentiating between healthy and cancerous tissues that a school of thought suggests human radiologists and other imaging specialists won’t exist within a few decades.

However, AI still has a long way to go, and even when our narrow AI systems are far more advanced than they are now, human intervention may still be required. As Ohad Oren et al conclude: “Unless AI algorithms are trained to distinguish between benign abnormalities and clinically meaningful lesions, better imaging sensitivity might come at the cost of increased false positives.”

Medical Adherence

Fred Keislinger, MD, writes in the NIH’s National Library of Medicine and the National Center for Biotechnology Information, “Medication nonadherence for patients with chronic diseases is extremely common, affecting as many as 40% to 50% of patients who are prescribed medications for management of chronic conditions such as diabetes or hypertension. This nonadherence to prescribed treatment is thought to cause at least 100,000 preventable deaths and $100 billion in preventable medical costs per year.”

What’s the point of new therapies or even high adoption rates if adherence or abandonment persists at high levels?

While the most productive vehicle for improving medication adherence is pharmacist-led interventions, “AI-assisted technologies have demonstrated increased adherence rates in patients using the technology vs. conventional care (30, 31, 33).”

MARKET BEHAVIOR ANALYSIS FOR COMMERCIAL AND MEDICAL SUCCESS

While traditional market research approaches can fall short in providing timely and actionable insights, integrating multiple data sources and using AI-supported analysis offers near real-time behavioral intelligence, empowering pharma teams to make informed decisions and stay ahead of the competition.

ChatGPT: right or wrong for pharma?

ChatGPT (GPT stands for “generative pre-trained transformer,” referring to how the chatbot processes requests and formulates responses) is an artificial intelligence chatbot that uses natural language processing to create conversational dialogue. It can respond to questions and compose written content like articles, social media posts, essays, and code. It’s available to the general public through OpenAI.

In healthcare-related settings, chatbots may be used to communicate with patients on behalf of healthcare service providers, to improve communication between doctors and patients, and to make healthcare more accessible. Applications for chatbots in these settings include mundane activities like scheduling appointments or ordering medical supplies to more complex operations like providing information about conditions or symptoms.