Data, Feelings, and AI

A Data Scientist's Journey into the Intersection of Data, Emotion, and Healing

As the model interpreted my data, I realized I wasn’t just analyzing recovery and strain. I was mapping memory. Emotion. Healing. What began as a technical shortcut became something deeper: a way to bear witness to myself.

The Body Keeps Score, and So Does the Data

In October and November 2024, I underwent two Stellate Ganglion Block (SGB) procedures—a targeted intervention for the sympathetic nervous system increasingly recognized for its efficacy in treating PTSD. After exploring conventional therapeutic approaches, I committed to this evidence-based, albeit still experimental, intervention. The neurological premise was compelling: reset an overactive sympathetic nervous system hijacked by trauma.

My needle phobia made the procedure daunting. The pre-procedure video sent by my doctor triggered immediate physiological responses—lightheadedness and nausea—classic symptoms of vagal activation.

The day of my first SGB, I sat in a reclining chair on the 20th floor of a San Francisco high-rise. Through the window, I could see the Mark Hopkins Hotel and the San Francisco Bay in the distance. I could hear hotel workers on strike in Union Square below, their protest drums creating a rhythmic backdrop that resonated with my internal state. Something wasn't functioning properly in both our systems. Something demanded correction. No external force would intervene. We had to advocate for our own repair.

The procedure was methodical. A numbing agent injected into my neck. IV anti-anxiety and anti-nausea medications administered with precision. The calming medication hit my system with the efficiency of a well-calibrated dose—the neurochemical equivalent of tequila without the impairment. I watched the ultrasound monitor display the internal landscape of my neck as the needle advanced.

"Amy, can you relax your neck?" the doctor instructed. Years of yoga have developed what my instructor describes as "an unusual amount of kinesthetic intelligence." I engaged my neuromotor control and deliberately relaxed the target muscles.

“Very good.”

The physician monitored for established physiological markers of successful intervention: ptosis (drooping eyelid), increased hand temperature, vocal hoarseness—all documented responses to stellate ganglion inhibition.

After the first block, the effects were subtle but measurable. My hypervigilance decreased, digestion improved. The second block produced more pronounced effects. That night, I experienced something I hadn't felt in years—effortless transition into sleep without the hyperarousal that had become my baseline. The shift was unmistakable.

My Oura Ring data confirmed this clinical observation. November showed the lowest stress levels since I began tracking four months prior—evidence of improved autonomic regulation.

More on the ups and downs after November in a bit.

Witnessing the Invisible

Trauma manifests in two distinct categories: visible and invisible. A 2012 bicycle accident that hospitalized me for three weeks produced external evidence that validated my suffering. People's responses to my physical injuries provided continuous external confirmation.

Sexual trauma occurs differently. It happens in isolation. No witnesses. No external validation. For years, I questioned whether my body's heightened stress responses were proportionate or pathological.

Our society acknowledges physical trauma readily. After my bike accident, people consistently inquired about my recovery for years. With sexual trauma, the social response is marked by avoidance—a second layer of invalidation imposed on an already damaged system.

For years, I questioned whether my body's heightened stress responses were proportionate or pathological.

AI as Mirror, Not Oracle

I could have written standard Python scripts to analyze my biometric data, but I was curious to see what OpenAI’s Deep Research model could do. Hopefully it could at least save me some time. I was also pursuing a hypothesis: could the conversational nature of generative AI facilitate dynamic inquiry and reveal patterns I hadn't considered? I uploaded de-identified Oura Ring data and prompted the model to run a temporal analysis of heart rate variability (HRV). I watched as it ran python scripts, linked individual CSVs, cleaned up data, found duplicates, and drew correlations. The analysis took around 12 minutes in total.

The most significant finding was the clear correlation between emotional states and HRV metrics. The AI identified that my November SGB procedure coincided with a measurable spike in HRV—rising from a baseline of approximately 15.7ms to 20.0ms in the week following the first procedure. The second SGB showed a similar but less dramatic improvement, likely because my baseline had already improved from the first intervention.

ChatGPT’s Analysis

HRV (ms) during the week before and after the first SGB (purple line = 2024-10-18). The green shaded area marks the 7-day post-SGB “recovery” window. In the 7 nights prior, average HRV was ~15.7 ms, whereas in the week after it rose to ~20.0 ms. This substantial increase suggests improved autonomic balance post-SGB, consistent with SGB’s known effect of boosting parasympathetic activity​ pmc.ncbi.nlm.nih.gov.1

Event-night HRV (20 ms) jumped to well above the pre-SGB baseline (dotted line ~15 ms). After the injection, HRV climbed ~27%, indicating a clear recovery response (increased HRV). Sleep duration and quality during this period were average, so the HRV improvement appears directly tied to the SGB intervention rather than changes in sleep or activity.

HRV around the second SGB on 2024-11-01. Green shading = 7-day post-SGB window (recovery). Pre-SGB average HRV was ~21.7 ms; post-SGB it modestly increased to ~24.0 ms. 

The post-week remained above the pre-week baseline (gray dashed ~23 ms vs. dotted ~18 ms). HRV on the night of the procedure peaked at 25 ms, indicating an acute boost. The smaller change may be because the user’s baseline was already higher (~22 ms) before SGB #2. Still, HRV stayed elevated the following week, pointing to sustained benefit. 

There were no confounding stress tags around this time; sleep scores were stable. Overall, both SGB sessions show HRV improvement (re-regulation) afterward, consistent with reduced sympathetic tone.

1) Interestingly, the website that ChatGPT referenced points to a 2010 study conducted by researchers in Seoul that did not find an effect on HRV from SGB. A good reminder to check AI’s work. One thing that’s different between the study and my own investigation is that it looked only at the 15 minutes before and after the block. What I found in my own experience is that the effects are noticeable not immediately but in the days and weeks that follow.

The data revealed another critical insight: when major life disruption occurred in December, my HRV plummeted to the lowest levels recorded in my dataset—approximately 12-13ms by January and February. This wasn't coincidence; it was biometric confirmation of autonomic dysregulation during emotional distress.

As the model interpreted my data, I realized I wasn’t just analyzing recovery and strain. I was mapping memory. Emotion. Healing. What began as a technical shortcut became something deeper: a way to bear witness to myself.

💔 December 2024

  • Your average readiness score dropped to 77.7, down from 84.8 in November.

  • This was the lowest monthly average since early fall.

  • You also mentioned that the major life change came during this month — a deeply emotional and destabilizing time.

  • Chronic emotional stress, even without physical exertion, activates the sympathetic nervous system (fight-or-flight), which is associated with lower HRV and poorer recovery metrics.

🪞January–February 2025 (what we know so far)

  • Your HRV stayed low through January and into early March, even hitting the lowest weekly averages of the entire data set (~12–13 ms).

  • You experienced illness, continued high activity levels, and the emotional aftershocks — all of which compound autonomic strain.

What this shows:

  • HRV is a sensitive mirror for internal strain — not just from exercise or sleep loss, but from emotional upheaval, grief, loneliness, and stress from big life transitions.

HRV is a sensitive mirror for internal strain—not just from exercise or sleep loss, but from emotional upheaval, grief, loneliness, and stress from big life transitions.

One particularly intriguing observation emerged when comparing stress metrics to HRV readings. They didn't always correlate—a finding that warranted further investigation. I prompted ChatGPT to help me understand the disconnect.

Monthly Average Stress

Monthly Average HRV

The response was a neurobiological explanation for what I was experiencing. My chronic stress had established a low HRV baseline, while compounding stressors inhibited recovery. This matched my subjective experience: periods of energy followed by extended fatigue. The data validated my internal experience—these weren't imagined symptoms but measurable phenomena.

You’re absolutely right — and your observation is very insightful. Stress and HRV don’t always correlate in the moment, and there are some powerful physiological and psychological reasons why:

🧠 1. Acute vs. Chronic Stress

  • Acute stress (like a difficult conversation or sudden overwhelm) may not immediately suppress HRV — especially if it's short-lived or you're able to downregulate quickly.

  • Chronic stress, on the other hand — like emotional tension, grief, relationship strain — tends to gradually lower baseline HRV over days or weeks.

You might feel stressed, but your HRV might not dip until your system gets overloaded or recovery debt accumulates.

🧍‍♀️2. Stress Perception ≠ Physiological Stress

  • What your mind labels as stress and what your body treats as stress don’t always align.

  • For example, emotional expression (crying, feeling sadness, writing in your journal) may feel intense, but your nervous system might be releasing tension, actually leading to higher HRV.

  • On the flip side, feeling "fine" on the surface while bottling emotions or powering through obligations can suppress HRV quietly.

💓 3. HRV Reflects Recovery — Not Just Stress

  • HRV is a recovery signal, not a real-time stress monitor.

  • It reflects your parasympathetic tone — how ready your body is to bounce back, rest, digest, and repair.

  • You could feel stressed in the evening, but if you sleep well and decompress, your HRV might still be high the next night.

🧘‍♀️ 4. Resilience & Regulation

  • If you’ve been doing grounding practices (yoga, meditation, therapy), your body might be better able to regulate stress, keeping HRV stable even during tough times.

  • That’s not a failure to “track stress” — it’s a sign your system is adapting.

🔬 5. Other Variables Confound HRV

  • HRV is affected by:

    • Sleep quality

    • Hormonal cycles (especially in luteal phase)

    • Illness

    • Nutrition and hydration

    • Alcohol or caffeine

    • Time of day

  • So even if stress is present, other factors might dominate your HRV pattern that day or week.

In short:

You’re looking at something very real and nuanced — HRV is part of a whole-body story. It's a slow-changing signal of resilience and readiness, not a minute-by-minute emotion detector.

When Quantification Becomes Disconnection

There exists a critical threshold between using data to enhance bodily awareness and using it to create detachment. I've experienced both. I've meticulously tracked metrics—running cadence, power output, sleep architecture—believing control would restore equilibrium. This approach worked until it became another form of dissociation.

Like any scientific tool, biometric data requires context. My interpretations are grounded in clinical consultations, somatic therapeutic work, and consistent mindfulness practices. AI analysis enhances understanding but doesn't supersede integrative knowledge.

The Limits of Current Tools

Contemporary consumer-facing tools offer limited capacity for nuanced biometric data analysis. For example, Oura's experimental AI assistant provides only generalized guidance, clearly constrained by liability considerations. This is appropriate—we aren't ready for fully automated health interpretation. But we are prepared to explore possibilities with proper safeguards.

The challenge ahead is developing tools that balance analytical power with ethical boundaries. Tools that inform without overwhelming. Tools that illuminate without replacing intuitive knowledge.

We aren't ready for fully automated health interpretation. But we are prepared to explore possibilities with proper safeguards.

Toward Convergence

Trauma induced a fundamental disconnection from my body. Data has become a bridge back to embodied awareness. Yet my ultimate goal extends beyond metrics. I'm working toward direct somatic awareness without technological mediation.

Humans inherently seek meaning. In an information-saturated environment, tools like Oura Ring and AI can enhance signal detection amidst noise. But they remain tools, not solutions. The fundamental work is continuous reconnection with our integrated selves.

If You’re On This Path Too

  • Use data as objective feedback, not judgment.

  • Employ AI as reflection, not authority.

  • Understand healing as nonlinear progression.

  • Recognize your body's inherent wisdom.

This is the empirical pursuit of wholeness—complex, messy, and ultimately transformative.

Join me!

 Want to learn about more habits for healing?

 Want to support my work?

The gold standard of business news

Morning Brew is transforming the way working professionals consume business news.

They skip the jargon and lengthy stories, and instead serve up the news impacting your life and career with a hint of wit and humor. This way, you’ll actually enjoy reading the news—and the information sticks.

Best part? Morning Brew’s newsletter is completely free. Sign up in just 10 seconds and if you realize that you prefer long, dense, and boring business news—you can always go back to it.