Can AI Emotion Analytics Enhance Personal Network Analytics?

Personal Network Analytics is the most systematic way to understand the connections we have with the people around us. These connections are a vital aspect of identity and human experience. It supports:

• Actionable quantitative and qualitative data we’ve underappreciated, providing a more holistic view of a person’s interpersonal ecosystem.   
• Engaging conversations and deeper probes with respondents, enhancing our ability to identify more of what they care about.
• Giving respondents something of deeply personal, tangible value, encouraging them to be more open and honest. 
• Displaying the bi-directional relationship between respondents and others within their universe of connections, providing deeper insights into influence opportunities.
• Identifying unmet needs, justifying new product and service offerings.
• Defining personas without engaging human respondents, enhancing digital personas. 

Is there a way to go deeper? Using AI, perhaps?

Our Hypothesis

Personal Network Analytics has enhanced the understanding of the life experiences of patients with rare diseases. Could adding AI deepen empathy even more, and if so, how? 

AI is ubiquitous, so where should we start? With literature reviews? Automated transcriptions? Transcript summaries?   

We opted to test AI Emotion Recognition to determine if it could identify how people with rare diseases felt towards their network connections. If this combination succeeded, we’d have new insights to inform R&D, healthcare professional (HCP) education, patient support programs, advocacy, and market access for all companies, not just those pursuing therapies for rare diseases. 

Rare Disease Patient Case Study

This case study describes the Personal Network and AI Emotion Analytics of a nearly 40-year-old woman, who was diagnosed with a rare disease as a teenager, and has been diagnosed with additional rare diseases since. She is married with several children and is a stay-at-home mom, not by choice, but due to her disease conditions. One, a condition that causes her constant pain, has yet to be diagnosed and cannot be treated. 

Partly to find support for her journey, she turned to social media, created a patient community, and became an influencer. Her plans for the future include expanding that effort. She admires other patients who visit Congressional Offices and aspires to become an advocate on Capitol Hill.

Personal Network Analytics Results

We focused only on her first-degree connections, which total 152, which is smaller than average for two reasons. First, her family rents a home, so it requires fewer connections to maintain or improve it. Second, her conditions limit her ability to engage socially. Her only personal social engagements are daily phone calls with friends from childhood who live in a distant state.

As the mother of three children, responsible for ensuring their safety and well-being, she has second-degree connections (e.g., teachers, doctors, dentists, and parents of her children’s friends). Her children play seasonal team sports, where she connects with coaches and other parents.  We estimate that her second-degree connections number 600. Whether first- or second-degree connections, each one has the potential to involve positive or negative emotions.

Glenna recorded her standard interview to collect first-degree connection data, developed the network map, and used it to structure a recorded debriefing session to gather the patient’s reactions.

AI Emotion Recognition Analytics

Sidi analyzed the recordings using Voice Emotion Recognition, an AI-based tool that identifies a speaker’s emotions through the tone and pitch of their voice, working equally well across languages, accents, and demographics. The feelings it detects can be subconscious and are not dependent on the words people use.  This highly robust tool predates Generative AI by over a decade, boasting accuracy rates of 85% to 90%.

We use a color-coding system in which vibrant green is the most positive emotion and vibrant red is the most negative. Other, less vibrant or yellow colors indicate where the emotion could be found along that spectrum. Network connections with no color indicate insufficient data to indicate an emotional valence.

AI Emotion Recognition Results

Her most positive emotions include:

• Two specialist physicians: a Neurologist, with whom she feels a strong partnership, and a Psychiatrist who prescribes medication for anxiety and depression (Health and Vitality Network).
• Deceased Paternal Grandparents (Ghost Network), which is consistent with recent studies indicating that children’s positive experiences with grandparents provide a “buffer” that helps them cope with difficult circumstances later.
• The fact that she has health insurance (Home and Personal Affairs Network).
• Anticipating her future as an advocate on Capitol Hill (Career Network).

Also positive, though less so, are:

• Role models, especially other patient advocates (Ghost Network).
• People in her online community (Career Network).
• The state of the family's legal, insurance, and financial affairs(Home and Personal Affairs Network).

She finds several connections challenging to manage:

• Five of her seven specialist physicians (Health and Vitality Network).
• Cleaning and caring for her household (Home and Personal Affairs Network).

Her emotions become somewhat negative when discussing:

• The challenges of remaining fit because exercise exacerbates pain (Health and Vitality Network).
• The general lack of support from others, including her mother, because her conditions are “hidden” and she does not “look like” she is in constant pain (Family Network).

Her most negative emotions include:

• In interviews, she described engaging in spiritual practices as a source of support, but Emotion Analytics revealed strong negative feelings, perhaps because she does this alone, adding to her feelings of isolation (Spiritual Network).   
• Though she feels positive about having health insurance, she feels strongly negative when she needs to use it (Home and Personal Affairs Network).
• In interviews, she describes a previous Neurologist as the most negative connection, and that is borne out in AI Emotion Analytics. He dismissed her symptoms and misdiagnosed her condition, making her feel ignored and worsening her condition over time (Ghost Network). At her request, we placed this and other past specialists in a group called “Specialist Care – Past,” because she intends to share her map with current healthcare providers (HCPs) to help them better understand her journey and any frustrations she may express as it continues (Health and Vitality Network).

The Verdict 

Did AI Emotion Analytics enhance Personal Network Analytics? We agree that it did.

Personal Network Analytics identified a variety of unmet practical needs. AI Emotion Analytics added nuances that an interviewer alone did not detect, specifically related to:

• The depth of her unmet emotional needs for support.
• Distinguishing between the positive emotion of having insurance and the very negative emotion associated with using it.
• Her fraught relationship with spirituality as she, unsuccessfully, turns to it for support.

Value for Marketers and Healthcare Professionals 

What types of patients or conditions might benefit most? Any that:

• Has a strong social-aversion component (e.g., disfiguring skin conditions).   
• Carries social stigma (e.g., depression, obesity)
• Impairs personal agency or makes interpersonal interactions difficult (e.g., hearing/vision/mobility loss, food allergies). 
• Is contagious, emotionally charged, or triggers guilt (e.g., STDs).
• Is lifestyle-related (e.g., some liver and lung conditions).
• Require help from navigators or therapists to manage (e.g., organ transplantation, cancer, degenerative diseases).
• Lack comprehensive patient education programs (e.g., most conditions).
• Need new advocacy messages that describe the burden of the illness (e.g., most conditions).