Introducing inVibe’s Topic Analysis Tool: Transforming Complex Voice Data into Actionable Insight Grounded in Real Human Emotion

What Is Topic Analysis?

At inVibe, we talk a lot about “what is said,” “how it’s said,” and “how it sounds.” What is said captures the content of the response, how it’s said is evaluated using discourse analysis, and how it sounds is measured using our machine learning models. Until recently, our language experts had to manually integrate these three aspects of the response to uncover “why it matters,” but now anyone using our dashboard—not just those of us with expert training—can use our new Topic Analysis tool to produce actionable insights.

Topic Analysis combines three core features of the inVibe platform to help any user quickly uncover the emotions associated with individual topics of interest: 1) computational analysis of emotion in the voice, 2) automatic detection and tagging of naturally emerging topics and themes, and 3) our validated large language model (LLM).

Without going into the details about how LDA topic modeling works, our model works like a literary detective that groups words often found together in and across responses into “topics.” For example, words like survival, reduction, and control might form a ‘Treatment Efficacy’ topic. Responses are composed of many topics in varying proportions as an individual naturally moves from one topic to the next.

What Can It Do for Us?

Topic Analysis allows us to examine the voice data to see patterns across respondents at a glance as well as to zoom in to uncover underlying patterns across topics. For example, let’s say we want to get a quick sense of HCP reactions to new data released at a major oncology conference. Topic Analysis identifies the key topics in HCP responses (e.g., ‘Efficacy of Targeted Therapy,’ ‘Toxicity Profile,’ and ‘Treatment Decision-making’) and aligns them with their associated emotionality in the voice (e.g., excitement, positivity, and boredom).

Just looking at the most discussed topics in reaction to these two data releases, we see pretty equal time spent discussing ‘Clinical Trial Data Impact’ but with very different emotional tones. HCPs are more optimistic about the Sabari study’s clinical impact than the Reckamp study’s impact, which they are more neutral toward. This leaves us with the obvious question “What’s driving this optimism?”

To begin to answer this question, we just click the topic of interest to have immediate access to all the responses where this topic is discussed. This page includes an AI Smart Summary of how HCPs in aggregate discuss the topic and a fully interactive chart that allows users to drill down into how it is discussed in individual responses.

If we want a more in-depth comparison between topics, we turn to the Compare Topics feature, which plots all discussions of one or more topics of interest within and/or across questions. When we click an individual response, the tool highlights which portions of the response align with that topic and situates them in their original context, which allows anyone to quickly see what’s driving differences in emotionality and how those differences manifest in the language used.

Upon inspection, we see that HCPs are more optimistic about the Sabari study because of the promising efficacy outcomes in a difficult-to-treat patient type while they are more neutral toward the Reckamp study because they view the data with greater uncertainty, viewing it as more preliminary and in need of confirmation. In just a few clicks, we provided the sponsor of the Sabari study with insight into the “why” behind HCPs’ reactions to their data and with the assurance that their competitor’s data reported in the Reckamp study is still years away from having an impact on treatment decision-making.

Topic Analysis transforms complex voice data into clear, actionable intelligence for anyone and empowers commercial and clinical teams to move beyond surface-level insights by revealing not just what people are talking about, but how they feel about it—enabling faster, more nuanced decision-making grounded in real human emotion.

Want to Learn More?

Are you interested in learning more about how our AI-enabled platform can make your qualitative research simpler, more systematic, and more scalable? Schedule a demo with us today and see for yourself!