Trending Applications of AI in Pharma.

The pharmaceutical industry is increasingly adopting AI and LLMs (like ChatGPT) to enhance efficiency, speed up R&D, assist in drug discovery, and revolutionize commercialization endeavors. Leading pharmaceutical firms are either acquiring or forging partnerships with AI enterprises, with estimated AI expenditures expected to hit $3 billion by 2025. We’ve compiled an exhaustive list of use cases, sourced from top-tier media articles.

In this blog, we explore the top 5 trending AI applications related to commercialization and clinical trials in Pharma. We’ve reviewed and analyzed recent pharma articles on AI, offering insights into the most influential applications Pharma is employing. The original articles are cited for those wanting a deeper dive into these use cases.

1. Real-time Medical Updates for Patients and Physicians & Improved Adherence

Through chatbots powered by LLMs, companies can deliver personalized health advice and address frequent questions regarding medications and treatments. This aids healthcare professionals in making informed decisions regarding treatment alternatives, subsequently enhancing patient outcomes. Moreover, LLMs can be instrumental in elevating patient education and adherence. 

For instance, Pfizer’s digital assistants facilitate easier access to medical information for both patients and healthcare professionals. Pfizer’s Medibot supplies HCPs with specific details (like stability of temperature-sensitive medicines), ensuring well-informed healthcare choices. Pfizer’s Fabi provides immediate answers on drug availability and information about Pfizer’s products, streamlining patients’ access to essential medical information.

2. Precision Marketing and Sales

By leveraging AI insights, pharma companies can make data-driven decisions on whom to target, where those targets are and what would be the most effective method of targeting them. 

Integrating AI in sales and marketing not only distinguishes these firms from their rivals but also guarantees that healthcare providers and patients obtain the most pertinent medical information, customized to their specific needs. This article from Forbes delves into how AI empowers corporations to refine sales tactics and craft personalized sales pitches for augmented efficacy.

3. Efficiency/Regulatory Modernization

In a sector known for stringent regulatory scrutiny, AI emerges as an invaluable asset, automating tasks and shaping standards. Pharma companies can refine regulatory intelligence monitoring by incorporating a semi-automated methodology. Traditionally, regulatory teams had to sift through agency websites manually for updates, a labor-intensive task. By adopting NLP and Large Language Models, this information collection and summarization can be semi-automated. Such amalgamation amplifies the team’s competencies, allowing them to pinpoint essential concerns, deadlines, events, and regulatory judgments with heightened efficiency, culminating in informed decision-making.

4. Pharmacovigilance and Drug Safety

AI-powered systems possess the capability to oversee and examine real-world data, pinpointing adverse drug responses and safety concerns. 

With continuous surveillance of healthcare databases, social media, and other platforms, 

AI can notify pharma corporations of potential issues. Through thorough analysis of components, effects, and outcomes, AI can guide these companies towards resolutions before making hefty investments in R&D

5. Patient Recruitment for Clinical Trials

The AI models can use past and current data to make performance predictions that could enable users to optimize clinical trial operations. 

For example, Novartis uses Nerve Live platform, which applies analytics and machine learning to clinical trial functions, such as enhancing resource management, determining the optimal location for drug trials, and recruiting the most suitable patients at the right time..

We would like to learn about your organization’s AI applications. Please write to us at [email protected].

Leveraging LLMs such as ChatGPT to discover hidden gems in Pharma data

The pharmaceutical industry possesses rich sources of unstructured data, including market research interview transcripts, inbound patient calls, sales rep notes, research articles, emails, and public web posts from physicians and patients. Exploring and extracting key insights from this data is a challenging and tedious task. Although techniques like topic modeling and LDA have been attempted in the past, the contextual nature of language used has made it difficult to extract meaningful insights.

Recent advancements in Large Language Models (LLMs) such as ChatGPT and BARD have significantly improved machines’ ability to understand context. By leveraging LLMs as foundational building blocks, we now have the capability to create powerful technology that can effectively summarize unstructured text.

As an example, we developed the SetuChat summarization engine, which collected ~500 tweets from physicians discussing immunotherapy. Using this engine, we extracted key discussion topics from the tweets and organized them in a tree structure, facilitating interactive exploration of the content. The tool empowers users to delve into the tree to discover essential points and even view the original raw comments interactively.

For a more detailed demonstration, please watch the video below showcasing how SetuChat effectively summarizes the tweets and facilitates interactive exploration of the extracted insights.

Application of Artificial Intelligence in Pharma Industry

The rapid emergence of AI with recent innovations in Large Language models (LLMs) such as ChatGPT is having a profound impact on various industries. This technological advancement is being likened to the breakthrough of electricity, which had a profound impact on various industries. This article will delve into AI use cases pertinent to diverse pharma stakeholders.

Commercial Analytics:

Leverage LLMs to gain real-time, comprehensive insights into online discussions and opinions among physicians and patients. Examples include:

 

  • Obtain specific physician perceptions about your drug compared to competitors using custom trained LLM

  • Utilize network graph analysis to identify and engage influential digital opinion leaders

  •           Identify influential  patients for advocacy, brand ambassadorship, and patient-centric       initiatives
  •            Analyze the impact of clinical trials, direct-to-consumer campaigns, conferences, and new  launches on sentiment and share of mentions for your drug and competitor drugs

 

  • Continuously monitor competitor websites for key messaging changes
  • Utilize extensive patient data, including electronic medical records, to predict patient volume for each physician
  • Generate conversation starter email digests email digests for Medical Science Liaisons (MSLs), enabling stronger connections with physicians during sales detailing
  •  

Marketing:

  • Identify unmet medical needs through patient and physician data analysis, enabling targeted content creation
  • Discover patients for rare diseases through social media discussions, facilitating online education and treatment outreach.

  • Enable customized marketing using LLM-powered bots to generate thousands of personalized messages

  • Develop marketing campaigns based on positive physician and patient discussions to enhance brand visibility and lead conversions

  • Automate content approval processes, reducing time and effort
  • Summarize lengthy videos/articles into digestible formats for swift review. Sanofi has implemented this to automate research papers review

Medical Affairs:

  •           Enhance medical education through interactive modules and personalized lessons, promoting effective knowledge retention and skill development

Other Applications:

 

  • Accurately detect potential adverse events and track adverse drug reactions (ADRs) from sources such as clinical trials, electronic health records, and social media. Johnson & Johnson used this to
  •            Monitor and forecast epidemic outbreaks for optimized logistics and supply chain managemet

For more information on implementing this technology/applications of AI within your organization, please contact [email protected]

Top 2 Things to know regarding OpenAI privacy rules before using GPT

OpenAI’s ChatGPT has been on an exponential growth curve since its launch & has shown the ability to handle a wide range of user queries, from writing emails to generating code, giving rise to numerous use cases. In view of Italy banning ChatGPT, after raising concerns about its privacy policy from a recent data breach and the legal basis for using personal data to train ChatGPT, it becomes important for users to keep in mind the privacy policy/privacy rules of OpenAI to protect their personal data. In this article, we wanted to highlight the top 2 things users need to keep in mind to ensure their personal data privacy.

  1. ChatGPT may collect user data for system improvement. ChatGPT has initially provided a pop-up to users before starting usage asking users not to share sensitive data & that it can be used for re-training the models.
    • SetuServ’s Solution: Users can opt out of having their content used to improve OpenAI services at any time by filling out this form. Also, given this data privacy challenge, it is currently safer to use ChatGPT on public data. Alternatively, you can anonymize the data at your end for any internal data you are looking to load into it.

2. Starting March 1st, 2023, Open AI will not use data submitted by customers via OpenAI API to train or improve the models unless the user explicitly decides to share the data with OpenAI.

    • The good news is that now OpenAI API can be used by companies over ChatGPT to access the GPT models to keep their data private. Reaching out to companies like “SetuServ” which have a thorough understanding of the intuition of LLMs, limitations of data privacy & security of LLMs, can help companies with better implementation of LLMs on their data while taking care of any data privacy issues

Plugins: A Game-Changing Upgrade That Will Revolutionize ChatGPT

  • In November 2022, ChatGPT launched, becoming hugely successful. It was able to gain 100 million users within 3 months, which was the fastest in history for any app
  • The main limitation of ChatGPT until now was that it is limited to its training data as of 2021 & limited use cases, while with the launch of ChatGPT Plugins, it will enable a wide range of use cases due to access to more recent data, which it can access through the Plugins
  • ChatGPT has released 3 plugins(Browsing, Code interpreter, Retrieval) by itself and also started giving developers early access for building 3rd party apps
  • Retrieval Plugin: Let ChatGPT access personal or organizational information sources (with permission)
    • It allows users to obtain the most relevant document snippets from their data sources, such as files, notes, emails, or public documentation, by asking questions or expressing needs in natural language
    • It can be hosted & deployed at the end of the company
    • Generates value since it can be used to ask questions from 1000’s of documents stored within a company database at once & retrieve the required information

Example of a Plugin being connected to the United Nations & asking questions

  • Third-Party Plugins: Access apps from ChatGPT & trigger actions through natural language
    • While starting a conversation on chat.openai.com, users can choose which third-party plugins they’d like to be enabled. The user can then enter natural language to trigger actions within these apps. Below is an example of a prompt from the announcement
      • User Prompt: Looking to eat vegan food in San Francisco this weekend. Could you get me one great restaurant suggestion for Saturday and a simple recipe for Sunday (just the ingredients)? Please calculate the calories for the recipe using Wolfram Alpha. Finally, order the ingredients on InstaCart.
      • Action: In this context, ChatGPT has integrated plugins from prominent services such as OpenTable, which enables users to book restaurant reservations. Additionally, utilizing an undisclosed recipe, the model can calculate the calorie count using the knowledge engine Wolfram Alpha and ordering the necessary ingredients from InstaCart, a popular retail delivery service. 
    • The Plugins would create an ecosystem of apps similar to google workspace helping users to use natural language to use any app in the world
  • How to access as a user
    • Users of ChatGPT plus need to join the waitlist here: https://openai.com/waitlist/plugins
    • ChatGPT is currently expanding the availability of plugins to both users and developers, with an initial focus on a select group of ChatGPT Plus users and developers. ChatGPT plans to gradually expand access to a larger audience in the future.

A new era of AI for education: GPT and its potential impact

GPT is revolutionizing the way children learn. Based on its capability to personalize & teach based on socratic method, it is soon to become an AI study partner for children across different edtech companies. Below are a few examples of ed-tech companies which have started using AI for teaching:

  • Khan Academy 
    • Khan Academy, a non-profit organization providing free education to everyone, anywhere, is launching Khanmigo, an AI-powered virtual tutor and classroom assistant using GPT-4. The AI assistant aims to help students with different levels and gaps in their learning and also support teachers to tailor their lessons. The GPT-4 model can understand freeform questions and prompts, allowing it to ask individualized questions for each student. Khan Academy plans to run the pilot program responsibly with a limited number of participants while inviting the public to join the waitlist. The organization hopes that GPT-4 will help students understand not just how to solve a problem but also the concepts behind it.
  • Quizlet
    • Quizlet is a widely-used learning platform worldwide, utilized by over 60 million students for studying, practicing, and mastering various subjects. Over the past three years, Quizlet has collaborated with OpenAI, utilizing GPT-3 for a range of applications, such as vocabulary learning and practice exams. With the release of the ChatGPT API, Quizlet is introducing Q-Chat, an AI tutor that adapts fully to students’ needs by presenting relevant study materials through an enjoyable chat experience, using adaptive questions.
  • Koalluh
    • Koalluh is using AI to generate stories based on certain selections the children can make. It solves the challenge of making reading engaging for children
  • SetuServ, with its rich expertise in NLP & LLM technologies, can implement LLMs within your ed-tech product. Please feel free to reach out to us at [email protected] to understand how we can be your LLM integration partner.

Impact of clinical studies on the Sentiment of drugs: A Case Study

 

A pharma client with a major oncology brand asked us to assess how a specific clinical study result for cHL disease affected the  HCP perceptions of their brand versus competitors worldwide.

 

 

 

 

 

The challenge for the client was setting up a system for global coverage of HCPs across various sources and harvesting and analyzing them in real-time:

SetuServ’s proprietary PharmaSignals platform identified ~934 global Hematology KOLs and HCPs with an active digital presence.

57% of the HCPs are from the USA, followed by 33% of HCPs from Europe. Sources such as Twitter, Semantic Scholar, PubMed, ClinicalTrials.gov, and Oncology media sources from the US such as OncLive, TargetedOncology, and from the EU like Haematologica are covered as part of the analysis.

Analyzing the collected data of the HCPs showed that, after the interim results of the clinical study, the net sentiment of client drugs decreased significantly because of the better efficacy outcomes of the competitor’s drugs compared to the client’s drugs. In contrast, the net sentiment of competitors increased marginally.

In 2021, the net sentiment of the client’s drug improved for two reasons:

  1. Physicians mentioned Interim results of clinical study less frequently because it showed that competitor drugs had better efficacy outcomes
  2. Other clinical studies demonstrated better efficacy of the client’s drug when used in combination with other drugs, which improved the sentiment towards the client’s drug

These factors reduced the sentiment gap between the client’s drug and its competitors.

PharmaSignals will be capable of performing comparable analyses for other clinical studies or events and investigating their impact on drugs in various therapeutic areas.

To get a proof of concept on how this solution can uncover rich, actionable insights for your drug/therapeutic area, please reach out to [email protected]

Unlocking the Power of Social Listening for Pharmaceutical Companies

We recently had the opportunity to work with a BioPharma client that specializes in acquiring molecules. The client approached us with a specific challenge: they needed a mechanism for understanding physician perspectives on certain molecules, their mechanisms of action, and targeted diseases. Specifically, the client was interested in gaining insights into physicians’ views on Menin-MLL and MLL-WDR5 and which of the two is preferred for treating Heme versus Solid Tumor.

To meet this challenge, we leveraged our expertise in social listening to create a system for collecting and analyzing online data, ultimately providing the client with valuable insights into physician perspectives. In this blog post, we will share the benefits for the client from understanding physician perspectives.

We will also delve into the challenges that the client faced in setting up a social listening system and how we overcame them. By the end of this post, you’ll have a better understanding of how social listening can help pharmaceutical companies make more informed decisions and gain a competitive advantage.

The challenge for the client was setting up a social listening system for collecting online data and analyzing the data to generate insights.

Our PharmaSignals product, which is powered by custom-built APIs from a wide range of sources, proved to be an effective solution for this challenge. By collecting nearly 1,800 publications and physician tweets, our AI models extracted key entities such as Mechanism of Action, Disease, and Molecules, along with topic/sentiment analysis to capture physicians’ opinions and sentiments.

Our machine learning-based noise filtering model further identified 110 relevant articles containing both Mechanisms of Action and Disease, effectively filtering out any irrelevant articles for the analysis.

Two sample physician posts with other entities classified, showcasing how this tool can be used to extract key information:

Through our product, we were able to determine that Menin-MLL was preferred for Heme, with higher sentiment for Heme cancer, while WDR5-MLL was preferred for Solid Tumors.

Moreover, we found that within Heme, AML was the highest-mentioned disease for Menin-MLL, while within Solid Tumor, Bladder cancer was the highest-mentioned disease for WDR5-MLL.

To learn more about how we can help your pharmaceutical company make more informed decisions, please reach out to [email protected].

Use Of Consumer Language for Efficient Ad Copies, Descriptions, and Titles of your Product Listing

Do you want to ensure your promotions resonate with your customers in your ad copies, titles, descriptions, and bullet points? It cannot be easy to know what will draw customers in. That’s why understanding “drivers and drainers” is key. Drivers are the factors that draw customers in. When you understand those key drivers, you can use them to your advantage in your promotional efforts. For example, if your product appeals to eco-conscious customers, emphasizing how it’s made with sustainable materials or great for the environment can be a great driver. On the other hand, drainers are the factors that can turn customers away. These are the elements you want to avoid in your marketing. If you’re trying to reach an eco-conscious audience, for example, promoting the product’s low price point might turn them off. By understanding the drivers and drainers in your target audience, you can create promotional materials that speak to them. This can be the difference between a successful campaign and one that falls flat. So take the time to research your customers and identify their drivers. Then, you’ll be well on crafting the perfect ad copy, titles, descriptions, bullets, consumer language, and ingredients.

Let’s assume you are selling a smartphone on amazon and would like to emphasize your smartphone’s performance. Let’s see how this can be done. First, we collected reviews from amazon.in for smartphones in the price range above Rs.1,00,000. Then, we ran these reviews through our state-of-the-art machine learning models to extract topics and run sentiment analysis. We got all the reviews containing mentions of the topics related to “performance” – “performance”, “product performance”, etc. Then we ran these models through our topic extraction framework again to learn what customers are talking about when they are talking about performance. The following stacked bar chart shows the most frequently spoken topics with performance.

These topics can be used in any content like ad copies and descriptions where the performance of the smartphone needs to be emphasized. As we pick these up from customers’ voices, these words are more likely to strike the right chords in the customers’ minds.

In another case, we looked at a specific smartphone listing – Apple iPhone 13 Pro Max (256GB) – Graphite. When we analyzed the reviews with our topic and sentiment lens, we found the following topic sentiment stacked bar chart

The above chart shows that the “weight” of the phone is primarily neutral and sometimes negatively perceived. There was no mention of the weight on the amazon product listing page. Looking at the reviews, it’s clear that people didn’t expect this and felt let down.

Handling the expectation of heavier weight via description or bullet points on the listing page can help avoid more such negative experiences for other customers.

These are simple examples. We can gain deeper insights by drilling into your topics and competitor topics to highlight your product’s advantages, etc. If this sounds interesting and helpful analysis for your products, please reach out to us by filling out the form below.

Fill in the details for more information

Qualitative Research for your new product launch

Running a business in the digital age means constantly innovating and creating new products. But before you launch, you need to ensure your target audience will be receptive to your product – or not.

 

There are tools out there that claim to be amazon product research tools – Jungle Scout, Helium10, etc. These tools are good at quantitative aspects like which products are running out of stock, the price, discount variations, etc. They lack the qualitative aspects which help you gain an edge over your competitors.

 

When you launch a new product, addressing customer dissatisfaction is a key factor. By understanding what customers are looking for, you can create a strategy to fill the gaps and differentiate yourselves from the competitors. One way to do this is to identify the drivers and drainers of customer satisfaction in a particular category. Drivers are features that customers look for and value when making a purchase decision. Drivers are what will make customers stay loyal to a company and buy from them again in the future. Drainers, on the other hand, are features that customers find unsatisfactory or do not find important when making a purchase decision. Drainers are more important when you need to find new opportunities. They help identify and understand the niche segment in the market that is unsatisfied with the current status quo. By understanding both drivers and drainers, you can identify the needs and wants of your customers and create a product or service that meets those needs and can capture that market share.

 

Let’s say you are looking to launch a new smartphone in a price range above Rs. 1,00,000. To determine the drivers and drainers, you must analyse some customer feedback. In this case, we are taking amazon reviews. From the reviews, you should examine the volume of mention of different topics in customer feedback and the sentiment associated with them. We extracted this data and ran it through our state-of-the-art machine learning models which extract topics and do sentiment analysis. Then we visualized this data using the stacked bar chart.

We can see that support and charging/battery are specific actionable topics that customers dislike more. Close to 50% of customers are talking negatively about these topics. Some examples:

 

Charging:

Support:

You can focus on providing better after-sales customer support and faster charging/ more reliable and larger battery for your new smartphone product to attract this unsatisfied customer segment. There are other non-specific topics like experience, value, etc which can be further analyzed to get more specific topics. While improving on these draining topics, your product needs to do well on smoothness, color, camera, display, design, and speed as these topics are liked in almost all products in this segment and hence they would be the minimum expectations from any new product in this segment.

 

This is a simple example. There is a lot more we can understand by drilling down into the topics to get subtopics and more granular insights. If this sounds interesting and helpful analysis for your products, please reach out to us by filling out the form below.

Fill in the details for more information