A presentation given at Computing in Cardiology 2023. Abstract (submitted to conference, with a slight edit) Introduction: Wearable photoplethysmography devices such as smartwatches are now being used to detect arrhythmias from inter-beat interval (IBI) measurements. However, photoplethysmogram (PPG) signals are highly susceptible to motion artifact, which could lead to erroneous IBIs. This study investigated whether simultaneous accelerometry signals could be used to determine when IBIs can be accurately measured from PPG signals. Methods: The PPG-DaLiA and WESAD datasets were used, which were collected during activities of daily living and mental stress tasks respectively. The datasets contain accelerometry and PPG signals measured at the wrist from 15 subjects using an Empatica E4, alongside chest ECG signals. IBIs were estimated from PPG signals using the high-performance ‘MSPTD’ beat detection algorithm. Reference IBIs were derived from ECG signals using two QRS detectors, and time-synchronised with PPG IBIs. IBIs were deemed accurate if the instantaneous heart rate (IHR) calculated from PPG-derived IBIs was within +/- 5bpm of the ECG-derived IHR. The level of movement was quantified using the mean absolute deviation (MAD) of the accelerometry signal over 5 second windows. Results: When the MAD was used to predict whether PPG-derived IBIs were accurate, it produced an area under the receiver operator curve of 0.78 on all data, 0.77 on activities of daily living, and 0.82 on mental stress tasks. Mean absolute errors in IHR of <5bpm were achieved when using the following MAD thresholds to exclude data during high levels of movement: 13.0, 17.0, and 11.0 milli-gravitational units respectively. These thresholds approximately correspond to previously reported cutoffs between sedentary behaviour and low-intensity activities. Conclusion: Accelerometry can be used to identify periods when IBIs can be accurately measured from PPG signals.