Exploring AI and harnessing its potential within global medical information

Our medical information team recently conducted a structured quantitative and qualitative survey of Medical Information, IT and management roles within various pharmaceutical companies to learn more about how our colleagues view AI and its potential use in medical information.

According to survey results, those who have adopted AI have seen 75% increased efficiency and 60% faster response times in automated tasks such as data analysis and quality monitoring. Cost savings were recognised, along with improved accuracy and customer experience.

Yet, there is still very slow movement towards actual implementation of AI in Medical Information services.

Survey results disclosed that despite 89% of respondents being familiar with AI technologies, only 44% are using AI within their medical operation. Of those using AI, 50% indicated that they work with it daily, 25% weekly and 25% rarely.

Many of the organisations tapping AI for their Medical Information operation are doing so through an outsourced partner, saving on internal investment costs.  However, concerns run high on the reliability and accountability of AI provided by third parties as well as data security and regulatory compliance.

Medical Information departments can find change complex and laborious due to compliance and regulatory factors as well as resourcing and financials. Implementing AI within Medical Information faces serious reservations resulting in many barriers to entry that must be addressed and changes made for the industry to move forward.

The most serious barrier is AI’s inherent cost and its hard-to-quantify financial justification return. There are exemptions.

For example, when implementing AI for large US medical information teams, the opportunity to achieve cost-effectiveness within a reasonable time frame is substantially higher than in Europe, MEA or JAPAC where the technology must be built out for many countries’ languages and local dialects serving small populations.

Variabilities in regulations across countries and regions further complicate cost concerns. These must be taken into account for medical information services to remain in compliance.

Regulatory agencies in every country are currently working to establish policy guidance on the use of AI within medical information, but the prospect is daunting for whoever is the first to cross that line to force the creation of policy.

As a result, by the time AI tools begin to show a meaningful financial return in the face of these regulatory and language requirements, the next generation technology is in play. Prior AI implementations are now out of date. It is a difficult cycle to address.
It is not surprising that data privacy concerns also pose a significant hurdle to AI implementations.

Company data is highly guarded throughout the industry and any chance it could become compromised was expressed as a significant barrier by 50% of survey respondents in their qualitative remarks, with 18% concerned even when implementing through a third party.

Therefore, it becomes important to overcome the many ‘unknown’ aspects AI represents to dispel these concerns, as evidenced by the fact that 70% of respondents who used AI personally had a positive attitude towards AI implementation in Medical Information.

The learning curve further complicates adoption. As in any new technology or update, there are challenges to overcome in utilisation. AI represents yet another level of technology with which one must familiarise only to face continual revisions.

Many organisations need initial education on what AI tools are even available, how they work and how they might benefit. Once implemented, AI tools require ongoing training and adaptation – challenges many Medical Information departments are not yet equipped to manage. A shortage of internal AI expertise and trainers further hampers adoption efforts.

One can’t leave a discussion on barriers to AI entry without discussing the elephant in the room: job security. 25% of survey respondents noted that they believed AI could replace human roles, particularly in routine tasks.

Certainly, as automation continues to advance, some current jobs could be in jeopardy. However, it is important to look at where humans are putting their efforts and determine if they are being used in the highest value positions.

If AI can free up team members to spend their time in a meaningful conversation, or addressing a complex case, or doing the research to put together a customised response for somebody, as opposed to an intake role of obtaining name, address and phone number, then AI is worth serious consideration.

Rather than replacing humans, AI can augment human roles by taking over repetitive tasks and freeing specialists to focus on the complex, high-value human-connection activities supporting Medical Information’s mission.

For many pharma organisations, partnering may be the best way to begin implementation of AI. The partner may be an outsourced provider of Medical Information services who has built-in AI capability, or it might be a technology company that can provide the bespoke solutions.

Such partners are valuable in that they have explored and implemented AI functions and have relevant learnings to share on what does and doesn’t work.

These partners can then leverage their experience and technology to offer flexible options that meet bespoke needs and timing. As a result, an AI neophyte can forego starting from scratch and thereby avoid the many stumbling blocks others have already encountered.

Per survey respondents, partnering with outsourced providers to tap AI was a practical solution when facing internal AI barriers. These partners were seen to help mitigate risk, absorb technology investment costs, maintain tools to justify continued use and have personnel already trained and ready to go.

Indeed, 74% of the survey respondents chose their Medical Information services outsourced partner specifically because they offered AI tools, with 53% of those outsourced providers using AI for data analysis and process automation.

Despite the fact that those working with these outsourced providers were allowed access to advanced AI tools without significant internal investment, concerns over data privacy and compliance remained, with 18% expressing such concerns. As reliability is proven over time, these concerns should dissipate.

Despite reservations, the industry recognises the need to move towards accepting AI technology to improve the ultimate customer experience.
Key questions are what parts of medical information should AI support? What is the best set-up? Does the department utilise it, either alone or with a partner, in a piecemeal fashion testing one country or one service?

Does it implement across all pertinent functions and regions? Each implementation must reflect the unique needs of the company and the healthcare professionals and patients it supports, complementing but not replacing compliance with existing regulatory frameworks.

There are some companies currently using AI throughout their medical information service and others just using it for intake. Ideally the goal is to thoughtfully integrate AI to support teams, thereby allowing employees to focus on high impact, meaningful work while automation handles routine tasks. 