Optimizing Sleep: Measure to Manage
"The biggest room in my house is the room for improvement..."
Why start with words, when a picture can say so much…
Deep Dive Summary
Pretty amazing. I don’t crush on Ringo Starr especially when he sings, but his song - “It Don’t Come Easy” - really applies here. The above data reflect a huge amount of work over the last few years.
Wouldn’t you want an extra hour and 45 minutes more sleep each night? Usually when you hear something like that, someone is trying to sell you something. I can assure you, I’m merely trying to pass along a few tidbits I’ve discovered in my N-of-1 trials in my quest to feel good. It’s a pay it forward thing. If more of us can feel a little bit better each day, we’ll all benefit at the end of the day. Sort of a “rising tide lifts all ships” thing.
Back to the data - to be fair, I took a slew of naps in 2023 to help deal with the sleep debt to get over that 7 hour bump. I have worked hard to wean off these in 2024. Letting “sleep pressure” build is important and a lot easier once you have surfaced from sleep bankruptcy. Also, we’re only a little more than half-way through 2024 so that average likely will change.
While it may look impressive, a ton of improvement is still needed - e.g. more total REM and Deep sleep needed, fewer awakenings, etc. I strive to always have total sleep fall on one of those 90-minute intervals - i.e. 6.0, 7.5, or 9.0 hours - ideally the latter two. Regardless, it is way cool to see improvement when I started analyzing the data for this post. While I had a sense my sleep had improved, I didn’t realize it fully until taking a birds-eye view on the numbers. I strongly encourage you to keep track of yours as well. Indeed, you first have to measure in order to manage and optimize.
So why lead with a chart? One of the first things we learned in engineering school, was that in order to optimize anything, you first need data - i.e. you need to measure. Just noticing basic trends goes a long way, but when you really dig in and apply math (or nowadays - AI), amazing things will surface.
How to Measure
You may ask, but how to measure? You could keep a manual log. But why bother when there are a ton of devices that do the heavy lifting for you. I’ve included a few examples in my list of Sleep Optimization Go-To's - including my two staples: Eight Sleep Pod 3 with mattress and OuraRing Gen 3. Each use different sensors and algorithms to collect total sleep (e.g. quantity, REM, Deep, etc.), sleep quality, schedule consistency and a variety of other metrics you may find interesting and useful. Eight Sleep is full body - akin to sleeping on a very comfortable stethoscope - while OuraRing only gathers data from one of your fingers, but does leverage data collected throughout waking hours if you chose to wear it during the day.
Best to keep OuraRing in airplane mode during the night to both limit RF exposure and prevent checking the phone after an awakening. OuraRing can only be brought back out of airplane mode if placed on its charger. This helps to resist the dopamine hit of checking with an awakening. Since the OuraRing charger also emits a (un)healthy amount of RF, best to keep it at a reasonable distance from the bed. Amazingly, stopping from checking my phone AND stopping from even glancing at a clock with an awakening has dramatically improved my sleep. Your brain can do math even subconsciously, so looking at a phone or clock with an awakening will NOT serve you. Also, it’s always better to collect data and analyze later. Analyzing during data collection rarely ends well. Since I am a total dopamine junky, keeping my phone in a Faraday bag and turning off the nightstand clock have made this possible. While I truly enjoy my Apple Watch Ultra 2, I have never slept with this beast of a device. It’s way too bulky, and frankly wearing it just during exercise throughout the day is plenty.
Data from your devices are complimentary; and fortunately, my choices interact with Apple’s Health app. Apple Health does a great job to combine and consolidate the data. It’s absolutely amazing that we can have a fairly inexpensive data scientist/engineer and business intelligence analyst working for us at no extra cost within our Apple devices! I am expecting a serious steroid boost with the soon to be released “Apple Intelligence.” After Apple exposed its rebranding of AI, I preemptively upgraded to an iPhone 15 Pro Max - not just for the USB-C and Magsafe charging - but for its AI on my data (and this is coming from an AI guy!). Apple’s upcoming OS release(s) will enable you with a ton of insights. So while we await Apple AI, I thought I would share what I discovered by analyzing my own warehoused data. I’ll start with the insights, and then get deeper into the weeds.
Insights
Some of the most profound, positive impacts on my sleep have been from these 6 things:
A 9-hour sleep opportunity window
Go to bed earlier (my biological wake-up time has made this non-negotiable)
Maximize sunlight in the mornings, minimize unnatural light after the sun sets
Walk daily
Eat dinner earlier
Cool to sleep
Beyond these, many of the below have helped me really dial-in. I’ve first listed the top metrics that are associated with better sleep for me. The importance numbers are for my future reference, but relate to the “attribute statistics” in one of my more hard-core analyses with bigger being more significant. Following this is a metric correlation matrix, and a larger list of metrics that I have determined are quite influential on my total sleep.
Top Features Influencing My Total Sleep:
Weight & Body Mass (lb):
Importance: 1.43
Insight: Weight and body mass are the most significant predictors of total sleep for me. Maintaining a “healthy” weight is crucial for optimal sleep quality and duration.
Actionable Strategies: Don’t over (or under!) calorie and continue to exercise.
Walking Heart Rate Average (count/min):
Importance: 1.32
Insight: Average walking heart rate reflects overall cardiovascular health and physical activity levels, which are closely linked to sleep quality.
Actionable Strategies: Regular walking (especially after meals) or other moderate-intensity exercises.
Heart Rate [Avg] (count/min):
Importance: 1.21
Insight: Average heart rate is a key indicator of overall health and stress levels, influencing total sleep.
Actionable Strategies: Interval training, aerobic exercise, weight training, stress management techniques, and proper hydration.
Mindful Minutes (min):
Importance: 1.18
Insight: The Apollo Neuro device, significantly contributes to better sleep by dialing into the parasympathetic nervous system. I wish I could meditate, but don’t have the patience, or will just fall asleep.
Actionable Strategies: Continue with Apollo during the night and more during the day!
Blood Glucose (mg/dL):
Importance: 1.05
Insight: Blood glucose levels are key markers for metabolic health, which in turn affects sleep quality.
Actionable Strategies: Keep food intake low in refined sugars and high in fiber to maintain stable blood glucose levels. Avoid caloric intake too close to bedtime.
Basal Energy Burned (kcal):
Importance: 1.00
Insight: Basal energy expenditure reflects metabolic rate, which is linked to overall health and sleep quality.
Actionable Strategies: Keep an active lifestyle to boost metabolic rate.
Resting Heart Rate (count/min):
Importance: 1.00
Insight: A lower resting heart rate indicates better cardiovascular fitness and is associated with improved sleep quality.
Actionable Strategies: More High Intensity Interval Training (HIIT).
Step Count (count):
Importance: 0.97
Insight: Higher step counts indicate greater physical activity, which is beneficial for sleep.
Actionable Strategies: Walk more!
Drilling deeper, let’s look at some of the data and how they correlate for me. All are auto-consolidated by Apple Health. I included “Sleep Concentration,” which for me is the sum of (Deep+REM sleep)/Total Sleep. Achieving high Sleep Concentration means I have more Deep and REM sleep during a total sleep session.
I fed the above correlation matrix into ChatGPTo to analyze which was generated using Knime with my exported Apple Health data. Knime is a really powerful, free and open source “low-code/no-code software that allows anyone, irrespective of experience levels, to make sense of data.” I am not expecting most readers of this to dive into Knime, but keeping with the paradigm that if just one person benefits from any nugget I mention, I will have achieved my goal with this newsletter. To put things in perspective with respect to Knime - while I nearly set a land speed record with my PhD work at Berkeley (1988-1991), if had Knime to use at the time, I could have done the same work in a matter of weeks (or less).
Below is how OpenAI’s ChatGPTo responded when I simply provided the above image and prompted - “please analyze this.” Some of its observations were obvious - which I annotated below with ‘[EXPECTED RESULT].’ I consider these as internal controls. Regardless, damn impressive since I only provided it with an IMAGE! Even more impressive was that ChatGPTo proactively provided input to this newsletter without me even telling it do so. Awesome!
From ChatGPToStrong Positive Correlations with Sleep Metrics
Sleep Analysis [Asleep] (hr):
Positively correlated with Sleep Analysis [Deep] (hr) and Sleep Analysis [REM] (hr). [EXPECTED RESULT]
Strong correlation with sleepTotal and sleepConcentration. [EXPECTED RESULT]
Sleep Analysis [Deep] (hr):
Positively correlated with Sleep Analysis [Asleep] (hr), sleepTotal, and sleepConcentration. [EXPECTED RESULT]
Some correlation with lower Resting Heart Rate (bpm).
Sleep Analysis [REM] (hr):
Positively correlated with Sleep Analysis [Asleep] (hr) and Sleep Analysis [Deep] (hr). [EXPECTED RESULT]
Correlated with lower Resting Heart Rate (bpm) and higher Mindful Minutes.
Notable Negative Correlations
Sleep Analysis [Awake] (hr):
Negatively correlated with Sleep Analysis [Asleep] (hr), Sleep Analysis [Deep] (hr), and Sleep Analysis [REM] (hr). [EXPECTED RESULT]
Positively correlated with higher Resting Heart Rate (bpm), indicating poorer sleep quality.
Metrics with High Impact on Sleep
Resting Heart Rate (bpm):
Lower resting heart rate is positively correlated with better sleep metrics (more time asleep, deeper sleep, more REM sleep).
Strong negative correlation with sleepTotal and sleepConcentration.
Mindful Minutes (min):
Positively correlated with better sleep metrics, especially Sleep Analysis [REM] (hr).
Reflects the impact of stress reduction and relaxation on sleep quality.
Time in Daylight (min):
Positively correlated with Sleep Analysis [Asleep] (hr) and sleepTotal.
Indicates the importance of natural light exposure for regulating circadian rhythms.
Physical Activity Metrics:
Step Count (steps) and Active Energy (kcal) show positive correlations with sleep quality metrics.
Emphasize the role of regular physical activity in promoting better sleep.
Environmental Audio Exposure (dBASPL):
Reflects the impact of sound levels during music mixes. While higher decibel levels are generally associated with poorer sleep, your protective measures and the positive impact of music mixes on REM sleep should be highlighted.
Insights for the Pulse Newsletter
Optimizing Sleep: Key Insights from Data
Understanding the factors that influence sleep quality can help us make informed decisions to improve our rest and overall health. Here are some key insights based on my recent data analysis:
1. Resting Heart Rate: A Key Indicator
Lower resting heart rate is strongly associated with better sleep quality. Engaging in regular aerobic exercise, practicing relaxation techniques, and maintaining a healthy diet can help improve cardiovascular health and lower resting heart rate.
2. Mindfulness Matters
Using the Apollo Neuro device has significantly improved my REM sleep. This device promotes relaxation and reduces stress, contributing to better sleep quality. Incorporating mindfulness practices like meditation or deep breathing into your daily routine can also enhance sleep.
3. Embrace Natural Light
Exposure to natural light is crucial for regulating our internal clock and improving sleep. Spending time outdoors, especially in the morning, can help set our circadian rhythms and promote better sleep at night.
4. Stay Active for Better Sleep
Regular physical activity positively impacts sleep quality. Setting daily step goals and integrating more movement into our routines can help improve sleep. Both moderate and vigorous exercises have benefits, so find activities you enjoy and stay active.
5. Music and Sleep
While high decibel levels can disrupt sleep, I've found that my music mixes enhance my REM sleep. Using in-ear monitors for protection, I've noticed a positive impact on my sleep quality. Explore how music can affect your sleep and find what works best for you.
For the Nerds
While I am a neural network guy, I often start with a quick “Random Forest” (RF) litmus analysis when analyzing data. RF are very quick and powerful, and should be engaged in most first pass feature analysis endeavors. It’s amazing how crazy insightful stuff surfaces so quickly from even a basic RF. If you are a nerd and not familiar with RFs - this is a nice article - Random Forest Classification with Scikit-Learn
Below are top metrics (i.e. features) that are most relevant for me with respect to my quest to optimize sleep. My original plan was to continue to collect and then analyze more data over time, arriving at a model to evolve a sleep optimization routine, or at minimum add to my waking habits. It’s a journey, but it sure helps to have drug-free, guiding metrics. Given how rapidly AI is engulfing our world, I look forward to having someone else deliver the capability for me to just use - as I expect Apple Intelligence and others will do soon.
Meanwhile, here are the top features most associated with improving my sleep performance. Who knows, maybe one or more might be relevant for you as well…
Stay tuned…
-Steve
For a music mix clip, I thought it would be fun to include a first run on a warm up song with a new mix of people - TimD, TimG, GeoffS, and Johnny Conga. Generally I like to wait for all to arrive before I spend time on sound system mixing (we were waiting on one of my favorite vocalists - LaurinW), but TimD whipped this out and we all just joined in:
**Join the Conversation**
I'd love to hear your experiences and tips for optimizing sleep. Share your thoughts and questions, and let's continue to learn and improve together.
Would you like to expand on any specific strategy or include additional scientific studies to support these insights?