Quantified Self Experiments / QS - Changelog (Feb 2024)

Here i will document changes i made to my QS framework and some new things i do.

Feb 2024. Stopped EOG due it doesnt add too much to EEG. Succesfully switched to silicone tape (no glue at all)  ag/agcl reusable electrodes for Shimmer ECG setup. ECG montage now being done in less than 3 minutes once a day. 5ch EEG montage now being done in 3.5 minutes and less than 1 minute to start recording. Switched from Ooler matress cooling system to Eight Sleep pod v3. Did a blood test and start writing article about bloodwork.

Jan 2024. Added EOG (bipolar Fp2 to E1) for OpenBCI montages. Started shaking Shimmer 3 and OpenBCI at same time when i go into bed, for future syncronization possible for ECG and EEG signals by using accelerometer data (from 25 Jan).

Dec 2023. Switched from ZMax to OpenBCI with headband montage (F7, F8, T4, O1, O2 with T3 as ref and Fpz as ground).

Nov 2023. ECG/HRV processing pipepline improvements: pchip interpolation for lost samples / hrv processing, smoothness priors detrending fixes, changed clean ecg and detect peaks methods to 'neurokit' and added 2-step peaks correction - firs t step is iterative method 'kubios' from neurokit and second step is to remove ectopic beats with method 'malik' from  hrvanalysis library. I wrote small python service for that.

Connected my 2 next gen Stryds to Stryd Duo.

Also i've realized that ZMax internal clock have drift, i wasnt synced it for a long time and drift was around 10 minutes - that means i cannot use my old data to compare it epoch by epoch versus Oura / Fitbit, for now i started measuring that drift and it seems like a 30-60s per day. From now i'm going to sync clocks for Shimmer & ZMax every evening before going to bed and start them simultaneously.

First version of ECG processing pipeline is published at Github.

Oct 2023. Started wearing Shimmer ECG for 24/7 HRV collection. Replaced Fitbit Charge 5 with Charge 6 (27 Oct). Tried Shimmer GSR+ for PPG and EDA but was unable to get RR from PPG (too noisy) and was not able to get any insights from EDA. Will do another try in a future.

Sept 2023. Dreem 2 stopped exporting data, switched back to ZMax.

May 2023. Switched to 32 key keyboard and ISRT typing layout (from QWERTY). Started logging each keypress and building heatmaps for confirm if ISRT is optimal in my case. Keylogging also will be used to track laptop activity time in a future.

April 2023. After ~30 days of collecting ZMax data - switched back to Dreem 2.

March 2023. Bought Hypnodyne ZMax (full version) and started building EEG / PPG processing pipeline for ZMax data (EEG, PPG, light, noise, temperature, respiration) by using neurokit (PPG) and yasa for sleep stage classification (EEG). My python expertise improved. Replaced Withing Body Cardio with new Withings Body Scan. Now during HRV standing test my hands holding ECG handle - its easy to hold so i would not expect significant impact on HRV values.

Jan 2023. Switched from Polar H10 to Movesense MD for morning standing HRV test (60 secs stabilisation period and then 60 secs measurement period). This is because Movesense MD provides raw ECG data with 512Hz sample frequency and allows full automation for data processing.