大型药企依靠人工智能掌握行业动态

A competitive advantage is necessary for success across industries, but maybe nowhere so much as pharmaceuticals, where companies spend millions of dollars and thousands of hours researching how to get their developments through clinical trials and onto the market before their competitors.

But they don’t do it alone.

Behind the top pharmaceutical companies, as well as smaller biotech firms, consulting agencies like Lifescience Dynamics provide third-party credibility from dozens of academic scholars and analysts and, more important, supply valuable tools to provide pharma companies with insights and recommendations to speed up the development of their products and gain FDA approval.

“Pharma is a data-driven business,” explains Hussein Jaafar, a senior consultant at Lifescience Dynamics, who has largely led the charge on the team’s adoption of artificial intelligence. “To be able to consult our clients, we need to have access to as much data as possible.”

The power from Lifescience Dynamics comes from its five main technology products, which incorporate elements of artificial intelligence—including machine learning, large language models, and generative AI—to compute large data sets, amass information, and make educated recommendations.

On average, it takes eight to 12 years to discover, develop, and ultimately launch a drug. Along the way, pharmaceutical teams make several decisions, often under “conflicting, limited, or patchy data,” explains Lifescience Dynamics founder and president Rafaat Rahmani. To minimize risk, pharma companies are required to seek third-party research firms to validate their data and decision-making. That’s why Rahmani, who previously worked for Eli Lilly and other health care consultancies, started Lifescience Dynamics two decades ago.

Until the past few years with the explosion of AI capabilities, many of this team’s tasks were still done by hand, amassing thousands of hours of labor each year each. With more than 130 clients that hail from the majority of the world’s top 20 pharmaceutical companies, that was a hefty task but also left more opportunities for human error, a major challenge for something as regulated as the pharma industry.

Now, with the assistance of AI, some tasks take just 10 minutes, and confidence in the task is often 100%. Though Rahmani has long considered Lifescience Dynamics a technology-savvy company, the real benefit of that mentality has shown in its use of AI.

The areas of business where Jaafar has seen the biggest impact are possibly less sexy but unparalleled in value to clients and his own team: data collection, data analysis, and data visualization. Critical to the pharmaceutical industry is the tracking of clinical trials, especially by competitors. Jaafar explains that the team used to have “giant” Excel spreadsheets that a team member would need to physically click through, read updates online, then update the sheet. In 2021, they rolled out a machine-learning model that does this for the team by pulling information automatically from online registries like clinicaltrials.gov and continuously adding updates. The live feed automation, he says, has been key to streamlining their processes and increasing their effectiveness in meeting client expectations.

Similarly, he spearheaded a project that scrapes valuable information about sessions and drug updates from the major medical industry conference. Many of these events draw in upwards of 70,000 people with sometimes more than 5,000 sessions. It was a beast for a team to consolidate and analyze data before AI; now, the Lifescience Dynamics model pulls abstracts and details automatically, even summarizing and recommending sessions for attendance.

The insights gathered by Lifescience Dynamics all live in a client portal, allowing clients at any time to log on for a full look at their competitive intelligence projects, clinical trial data, and drug data. Jaafar explains that they are currently building AI models on top of that data to help clients query using natural language better understand the results. It not only adds transparency in the client-consultant relationship, but saves the Lifescience team from fielding time-intensive, resource-intensive questions from their clients.

More recently, Jaafar and his team looked at the benefits of generative AI, specifically around online surveys built to allow independent physicians to weigh in with critiques and recommendations for a particular drug. An important component of the peer review process, pharmaceutical companies reach out to physicians for real-world, patient-facing opinions on potential drugs. For Jaafar, generative AI and large-language models have allowed him to produce survey templates for online discussions among physicians as well as identify relevant experts for a specific survey.

“This was previously done entirely manually and we would just have to use our own experience and expertise to pull something together,” Jaafar says. “But with AI, we’re able to give it the background of the discussion guide we’d like to have, and it produces a very useful template that has us 80% of the way to a finalized guide.”

The team manually works on the remaining 20%.

While the team celebrates the success they have had with AI, Jaafar and Rahmani know bigger challenges await. Jaafar would like to build their own models for AI specific to their craft. Though Lifescience Dynamics can pull from its own historical data, the real value would come in more shared data from the industry. Unfortunately, he explains, the regulatory nature of health care and patient confidentiality combined with the competitive nature of the pharmaceutical industry means companies hold their own data close for a variety of reasons. A fear is that companies will continue to silo in fields of development rather than share collective data globally so that AI can learn at an exponential rate. There is simply less shareable data than other fields.

Rahmani predicts it will take more years to settle debates in pharmaceuticals over AI. For all the euphoria and excitement, there are old promises and leaders who just aren’t for technology, he says. He, however, feels confident in the future of AI as a tool to the industry’s collective success.

“I can understand why they aren’t willing to connect, but it limits the utility of AI,” Rahmani says. “Our clients engage us to give them the insight and convert insight into foresight, in the shortest time possible and in the least expensive way. These AI tools squeeze the most out of our data and bring that data alive.”