AI-Powered Social Media Research Reveals Patients Needs in Novartis Study

Michael Seymour |
 12/02/22 |
6 min read

AI-Powered Social Media Research Reveals Patients Needs in Novartis Study

Novartis is a multi-national healthcare organization based in Switzerland that uses science and technology to address some of the toughest health challenges in the world.

With a primary focus on research and development, the company is always discovering and developing breakthrough treatments and efficiently delivering them to patients.

Recently, the organization used NetBase Quid® for yet another breakthrough: Understanding the educational and emotional needs of people living with myelodysplastic syndromes (MDS).

Through consumer research centered on social media listening, Novartis has found and published a wealth of information to guide it and other healthcare providers (HCPs) in improving the quality of life for people living with MDS.

It is a great demonstration of how modern HCPs can make use of social media to reach and impact the lives of patients all over the world, subsequently improving the industry.

The Challenge in Assessing Patient Needs

The biggest challenge for people living with chronic conditions is usually learning a new way of life and adjusting to the physical limitations. The emotional burden as well as information gaps can severely affect the patient’s outlook and ruin the chances of treatment success.

In particular, MDS – a group of blood cancers – can be quite debilitating with its heavy symptom burden and limited options for treatment. The main challenge for caregivers is assessing the patients’ needs as the disease progresses.

Patients need a range of information on how to manage the disease, not least how to navigate through life. They also need emotional support. Moreover, these needs change across the different phases of the disease.

Hence, the education as well as emotional support has to be updated and catered to the patient at any particular stage.

Often, the primary sources of care for the patient – i.e. physicians – are not in a position to provide adequate information or emotional support. Patients often don’t share all the details of their daily experiences with their physicians, preferring to share with other patients.

Additionally, HCPs haven’t had the means to eavesdrop on patient conversations for insights on how to help. The alternative route for the latter is to consult online sources including social media networks and forums.

Paradoxically, these online sources hold a vast amount of data about people’s concerns and informational gaps that could help physicians offer adequate education and support to their patients.

Understanding the perspective of the patient on how MDS impacts their life is a necessary step to achieving a high level of care. For instance, what is the impact of the disease on people across different cultures and healthcare settings?

The answers lie within the shared experiences of different patients. Even better, the data is often readily available for study. Analyzing patient posts on the platforms can reveal unbiased common themes in the conversations.

Unfortunately, it is all text-based and thus unstructured. Hence, HCPs using traditional approaches to consumer research that rely on data being structured are still not able to understand their patients.

Thankfully, modern AI-powered technology is up for the challenge: Machine learning (ML) and natural language processing (NLP) for analyzing human speech.

Naturally, in its quest for better care for people living with MDS, one of the leading research and development organizations in the world turned to the most advanced techniques.

Novartis analyzed posts on online patient forums to understand why patients and caregivers engaged in online discussions. The study, conducted over six countries, was aimed at revealing the needs of patients and caregivers through the different stages of the condition: diagnosis, management, and treatment.

Using Social Media Listening to Close the Gap

Using NetBase Quid’s proprietary consumer research technologies, Novartis analyzed 20,000+ forum posts from patients in Canada, China, France, Spain, the UK, and the US.

The countries were chosen for having sufficient activity and eight forums were picked on the same basis. The posts were filtered to align with the introduction of HMAs – effective treatment options for MDS – while ensuring an adequate sample size. Thus, the study chose posts from 2011 to 2019.

Through NLP technology – which can analyze human speech and categorize it by the words used and semantic context – the study identified common topics (themes) in online conversations. A topic was determined by the common keywords appearing on multiple posts and was referred to as the motivation for engaging online.

While posts could often reflect multiple motivations, the study categorized them by the main motivation to allow quantification.

Beyond offering a simple tally of occurrences for each topic, the algorithm identified linkages between posts to show how connected the different topics were. This is how a network map is created on NetBase Quid to help researchers parse and understand the whole conversation.

How to read a network map of topics on NetBase Quid.

How to read a network map of topics on NetBase Quid.

To understand why patients were online at different stages of MDS, the posts were filtered along the keywords suggesting disease phase. Then further analysis revealed common topics. As NetBase Quid only accesses publicly available data without infringing on user privacy, firewall restrictions prevented analysis of posts from China.

Next, NLP sentiment analysis was used to identify positivity and negativity in posts. This would help to better understand the motivations of patients at different phases of MDS.

Positivity was indicated by words like “better” and “improving” while others such as “painful” and “disappointed” indicated negativity. Some were classified as neutral.

Additionally, the sentiment was quantified by assigning a score to each post on a scale of -100 to 100. To ascertain the accuracy of NetBase Quid’s NLP sentiment classifier, an independent classification done by humans was conducted on a number of posts.

Discoveries and Conclusion

Of the 20,000 posts analyzed during the study, 15,000 were from the USA. Across all six countries, patients had a range of terms to refer to the condition. The study was able to identify the distribution of common terms for each country.

The study made the following important discoveries about MDS patients:

  • Motivation to engage online: There are seven main motivations for people living with MDS to engage online. The study identified them as clinical, diet and lifestyle, education and logistics, emotional, physical, transplants, and treatments.
Motivation network maps for the different countries.

Motivation network maps for the different countries.

  • Online engagement activity by disease phase: People living with MDS may be more motivated during certain phases of the condition than others. The study observed high online engagement during the phases of diagnosis, treatment initiation, and after experiencing the treatment.

Sentiment analysis was also conducted to understand the relationship between disease phase and patient sentiment.

This analysis identified existing information and unmet emotional needs among people living with MDS. Specifically, the most common challenges were about emotional concerns, understanding the condition, and the available treatment options.

Differences in motivations and the types of patient needs could be explained by cultural differences and access to care. For instance, people in European countries were found to be more motivated by their emotional needs while those in Canada, China, and the USA sought to understand the disease.

Educational needs also varied at different phases of MDS and with social media listening, the study discovered gaps all through the patient journey with very specific demands such as the criteria for diagnosis, interpreting blood counts during treatment, and treatment options after one set fails.

Similarly, emotional needs were observed at every stage of the patient care pathway.

This information can help HCPs to improve patient care taking into account what different patients need. For instance, targeted support could be provided at the time when patients are most likely to need it e.g. immediately after getting the diagnosis.

Spurred by COVID-19, the shift to online support by patients should be embraced and harnessed for better care.

The study concludes that treatment developers and medical education programs should take advantage of the increasing use of virtual communication. It will help them better understand the needs of the MDS patient, inform patient education and support programs, and gauge awareness of treatments among patients.

By understanding the motivations behind MDS patient online activity, HCPs can develop strategies to deliver relevant content to patients where they can see it right when they need it.

In the end, such efforts by HCPs could encourage people living with MDS to play a more active role in managing the condition. They will also know what to expect living with it and develop a support network that will supply their emotional needs as they arise.

Comparing this study to a previous one done manually, the researcher says that this one analyzed a larger and more diverse sample and understood the patients better. Not only did it help validate previous findings, it also added to and elaborated upon them. Additionally, the current study was much less prone to bias.

This is the power of NetBase Quid’s next-generation proprietary AI technology. If you would like to obtain actionable results from your own consumer research, reach out for demo today!

Premier social media analytics platform

Expand your social platform with LexisNexis news media

Power of social analytics for your entire team

Media analytics and market intelligence platform

Enrich your media analytics with social data

Social media benchmarking
and competitive intelligence

Data streams & custom KPIs for advanced data science

AI, Image Analytics, Reporting Tools & more

Out-of-the-box integration with other data sources