While you’re paying attention to your client and making observations, Upheal is busy making some additional observations for you in the background too! Our team of NLP, psychotherapy and coaching experts has developed a way to measure important speech-based metrics using the session transcript.
​
Our goal is to help healing professionals gather as many valuable insights as possible. And while our team includes coaches and therapists, you are the wisdom and the experience. Ultimately, interpreting session analytics is up to you as you carefully consider their meaning in the context of every unique individual with due consideration of their case history.
​
Now, with that being said, below are some of the exciting ways we believe some of these measures can help you.

You’ll see it right underneath your session analytics. The session map shows speech cadence, the client’s moments of silence, sentiment and tense over time – in other words, over the course of an entire session.

A resemblance or symmetry between you and a client in terms of speech cadence may indicate good things! Research into LSM (Language Style Matching) shows higher LSM indicates a heightened level of attunement to the other (Pennebaker, 2011). Therefore, it may be one of the predictors of the strength of a therapeutic alliance, and potentially, therapeutic success. (Borelli et al., 2019.) In the session map, we've visualized your cadence alongside a client's. Speech cadence then, may indicate a client's ability to access aspects of their experience in therapy and aid in the development of the clinical process to convey information about an internal state and the therapist's attunement ability. (Rocco et al., 2018).
​
​Tip: Another way to use this measure is to connect any notable changes in cadence to the topics discussed in your session! Notice that in the image above that there are numbered tags just above the cadence section – these correspond to topic titles from your session transcript.

And finally, changes in cadence can also reflect mood state changes such as mania or depression, that for example, accompany borderline disorder. (Notably: pressured speech and poverty of speech.) Eventually, we'd like to be able to predict a client’s oncoming episode by looking at speech cadence over time. But that's probably too ambitious! 😅 For now, you can at least use this session by session.
​

Why this may be important: there is some evidence that in bad sessions therapists over-participated compared to their clients or acted very passively – while in good sessions, the client-therapist participation levels were more balanced. (Friedlander, 1985). Using the session map, you can see where you may have dominated compared to the client and vice versa, especially in relation to topics discussed.
​
And, as above, talking ratio may connect to LSM (Language Style Matching), and indicate a certain level of attunement or strength of therapeutic relationship, and potentially, therapeutic success. (Borelli et al., 2019.)
​
​

We define silence as the gap between a client’s two words or two sentences. (Note: it is measured only after passing the threshold value of 3-5 / 5-10 seconds or more (Cook, 1964). As with speech cadence, we recommend connecting moments of silence to the topics discussed. (Each number tag indicates a topic title from your session transcript.) By connecting moments of silence to specific topics you will see what prompted moments of self-reflection, deeper thinking, or an aha moment in your client. Moments of silence may also indicate that the process of verbalizing thoughts has moved to inner monologue or that a strong emotion has occurred. Potentially, it may also reveal moments of trauma response – when the client is in freeze mode for instance. In future, if we develop the ability to analyze this across sessions, it may do even more, such as reflect the growing trust in a therapeutic relationship.

​

We identify the number of sentences with Positive, Neutral or Negative sentiment using something called a polarity value. The aim of polarity detection is to find out whether the opinion expressed in a text is positive or negative in how it connects to the topics discussed. We calculate the polarity value using the Ascent model for English. On your session map, you can see how client sentiment changed over a session and in connection to the topics discussed. (Positive sentiment is green and negative is visualized in red.) This may help show where a client may be engaging in cognitive distortions such as catastrophizing or indicate low mood or depression if sentiment is continuously negative.

​
​

This measure identifies the number of words that are verbs in the Past, Present, or Future tense. The tense analysis is helpful for identifying where a client is stuck in their thinking, especially in connection to the topics discussed. Keeping track of tense allows you to see how long a client spent in the past throughout your session – connected to say – their childhood, which can indicate trauma and show past-driven processing. It can also help uncover time spent in future-based anxieties and/or indicate fantasizing and obsession. In contrast, more present-based speech and tense forms may mean a client with higher levels of mindfulness and the ability to connect to their emotions in the now (but we don't measure this yet).
​
​Is the session map helpful? We're introducing AI into the world of mental health to help healing professionals like you. Your feedback and ideas are essential. 🙏