The engaging session generated lots of important questions from the attendees. Here are some of the key questions and answers from the webinar.
[Time stamp: 7 min 46 sec] Question 1: Isn’t the algorithm setup by the user inputting the social query?
Jackie’s answer: “Some of the algorithms for sentiment are hardcoded into social listening platforms so that would be something that you would want to check when you’re evaluating the different platforms. So using Plutchik’s Wheel of Emotions we can align our categorisation and while we’re coding the conversations that we’re having according to Plutchik’s model the opposite side of the wheel is the contrasting emotion. So sadness is the opposite of joy, for example. So what does this mean for the patient’s emotional journey? Well, taking that kind of differentiation between the quantitative view versus the more qualitative emotional analysis here on the left hand side looking at the MS patient experience in the United States, if we look at the quantitative kind of automatically applied sentiment 54% of the conversation that patients are having is neutral okay whereas positive and negative are a little more fairly balanced over here but that doesn’t help you understand where you can address unmet needs of patients or understand drivers or barriers that they may have towards adoption or switching behaviour. But if you start to look over here on the right-hand side at the emotions that are being expressed, you can across the entire patient journey understand that there’s a lot of frustration happening here and while there is some neutrality in the conversations there’s also a lot of sadness and worry being expressed and so by understanding these just at this high level you can understand that there are already areas to dig into to identify unmet needs.”
[12:47] Q2: How are you identifying emotional, physical impact on quality-of-life? Are you working with MS experts?
Jackie’s answer: “These are all from online conversations and these are categorised into the different quality of life impact categories. So, as an example illustrating some of this, so patients are often frustrated when seeking diagnosis. You can see that the pre-diagnosis stage where they’re discussing all of their symptoms, and things like that, is, while it was lower on posts, there’s a lot of frustration, worry, sadness, because they’re not quite sure what’s going on and as an example of this you can see, coming from Reddit, it says it’s tough because ‘my doctor initially doubted me when I suggested it but he also kind of half listened to symptoms explained in appointments’. So you can see there’s some real frustration here that they’re not being listened to that diagnoses aren’t coming and you know their needs aren’t being met or addressed.”
[13:46] Q3: How does this work across multiple markets and languages as emotions are perceived and expressed differently and impacted by cultural nuances?
Jackie’s answer: “When we are setting up this research and starting to align on the coding framework, we sit down as a team who’s doing the coding and discuss how these emotions are categorised in conversations from each of the markets. We ensure we are aligned on what is frustration in this context, and in this cultural context, and language context, and align on that before engaging in the manual coding. We also have multiple checkpoints where we come back in and make sure that everybody is still aligned on that and discuss anything someone has questions about.”
[25:30] Q4: Why is understanding the patient, caregiver and HCP emotions important?
Jackie’s answer: “It’s important because if you ask people you know what the sentiment around this is, as you saw more than half of it was neutral. Well, that doesn’t help you understand anything about adoption of new treatments or unmet needs. A lot of people would be neutral or negative about this, but by going beyond the sentiment to the specific emotions you identify opportunity areas or drivers or barriers to adoption adherence switch content and really start to understand the needs of the patients and identify specific points in time and specific needs to address at those points in time.”
[26:40] Q5: How is emotional journey analysis different from automated sentiment detection?
Jackie’s answer: “The automated sentiment detection applies the same algorithm whether you’re talking about your watch or you’re talking about your MS treatment. So if something is ‘the bomb’, that could be taken as negative or positive but it’s going to be scored as negative or positive, the same every time, despite the context of the conversation. So this automated sentiment, it doesn’t understand that you’re speaking in a health context and some of these things have nuance that is very important to recognise. That nuance can be detected by going beyond this automated sentiment and doing the manual coding and understanding the context as well as the cultural relevancy.”
