Maximising the potential of AI in pharma

We know AI has the potential to improve healthcare efficiency and this is widely discussed both in the media and within industry. While realistic expectations about the potential of AI solutions are required – such as the risks of misuse, and how it will be integrated – the ethical implications of slow progress or failing to adopt technologies must be deeply considered.

Indeed, there are moral consequences of not adopting AI, or of failing to use it to its full advantage – namely, missed opportunities for patients to feel better and for treatment to be improved, and for more people to access good care.

There’s a growing appetite for improving the accessibility and availability of this technology from both consumers and businesses, as well as healthcare professionals (HCPs). Increased investment and innovation in AI health solutions specifically show recognition of its potential to address unmet clinical needs, and practical applications for AI are already happening in pharma with proven results. In 2024, 40% of all digital health funding went to AI-driven start-ups, a rise from 33% the previous year.

In many cases of AI and integration, we’re beyond pilot stages. This is a crucial inflection point to accelerate and realise technology – go all in – and overcome any hesitancies to scale effectively.

True medical and scientific progress, especially within research and development (R&D), involves constant refinement and innovation, and a natural step is to adopt and test new technologies in these scenarios. R&D is well suited to utilise machine learning and data-driven approaches, where there are large data sets and complex patterns to discover. AI can expedite the development of life-saving treatments, cutting the development timelines by 50%, according to the Information Technology and Innovation Foundation, and can improve pharmaceutical R&D, leading to personalised medicine approaches and more efficient drug discovery.

‘AI-driven personalised cancer therapies have led to a 40% improvement in treatment response and a 30% decrease in toxic side effects’

There is emerging evidence that AI can outperform human clinicians in certain diagnostic areas such as breast cancer, where AI diagnostic tools have improved the accuracy of diagnosis by up to 13% by analysing subtle differences in medical scans. AI-driven personalised cancer therapies have led to a 40% improvement in treatment response and a 30% decrease in toxic side effects. The use of AI technology by HCPs is critical, paired with widespread training, which pharma has an important role in facilitating.

The way we approach healthcare is changing. Now, people expect and need different types of care, so there’s no question that pharma needs to align with our tech-first and tech-enabled world. Without this shift, pharma will be stuck in a stifling continuity bias, using the same methods and not considering changing conditions. Disruption is one of the healthiest ways to effect change and AI can help support fresh thinking and new perspectives – defining the outcomes of what we want healthcare to look like will be important.

Let us take clinical trials and drug discovery as an example. AI can offer a new way of approaching clinical trials and has already demonstrated end-to-end improvements in the process:

• In the early stages of drug discovery: AI can support target identification, drug design and the analysis of genomics data sets
• During the preclinical trial planning phase: machine learning models can simulate different trial designs to predict potential outcomes, creating ‘synthetic’ arms of studies for comparison. It can also help identify the right patient to participate and ensure diversity of trial population
• In the post-clinical trial analysis stage: AI can help identify subgroups of patients who respond particularly well or poorly to a treatment, enabling a more tailored patient-centric approach.

AI is a tool that can empower people to take control of their own health, giving only benefits for long-term prevention and treatment of disease. When patients have more control over their health information, treatment plans and communication with clinicians, there’s a better understanding of (and satisfaction with) care.

AI technologies and digital health tools are increasingly recognised as key enablers of a patient-first experience:

• For patients: AI-powered chatbots and virtual assistants can streamline communication by answering patient questions, scheduling appointments and providing medication reminders. This reduces the administrative burden on HCPs and enhances patient engagement
• For clinicians: AI can alleviate administrative tasks or streamline patient monitoring for clinicians, freeing them to spend more face-to-face time with patients and focus on urgent cases. This improves the quality of care and strengthens the patient-clinician relationship
• For medical data analysis: AI algorithms can analyse vast amounts of patient data to create personalised treatment plans, optimise outcomes and significantly improve the overall patient experience.

We’ve established that AI has the potential to support efficiency: streamlining pharma R&D, enhancing production and improving distribution. With this in mind, it’s important to consider the potential this has to close the gap in health equity. What could realising the potential of AI mean for already underserved communities?

Underserved communities, for example, in lower-income countries may experience issues with healthcare access, since these communities can be remote and harder to reach. AI tools and telemedicine can help people access care – automating, scanning, alerting and connecting people through digital interventions. Just one example of this is m-mama, an AI-driven emergency referral system in sub-Saharan Africa that has reduced maternal deaths by 38% by connecting pregnant women with transportation to healthcare facilities.

Underserved communities often wait years to access new medicines. The use of AI in pharma can expedite drug development and manufacturing, so novel medicines get to underserved communities faster. Additionally, AI can integrate data across multiple complex data sets and settings, enabling bespoke tech solutions to address specific challenges at a global or local level.

Furthermore, AI’s ability to be trained on data sets specific to certain diseases or patient populations within particular communities – such as sickle cell disease, which disproportionately affects black people – enables the development of drugs tailored to specific population demographics.

AI can expedite drug development and personalise treatments, leading to improved patient outcomes and reduced side effects. It streamlines workflows, freeing clinicians for more patient interaction and enabling better access to care, especially in underserved communities. AI can also empower patients to manage their health and provide data-driven insights for optimised healthcare delivery.

While legitimate concerns about privacy, bias, accountability and access must be considered, the concerns over AI’s deployment should not overshadow its immense potential to level the playing field. By realising AI’s potential while protecting patient welfare, we can create a new era of data-driven, personalised healthcare that enhances clinical outcomes and healthcare access for all.

References are available on request.