Electric heat pumps feature prominently in projected energy transitions in the UK and elsewhere. Owing to their high electricity consumption, heat pumps are viewed as important targets for demand-side response (DSR). Findings are presented from a field trial of a new control system that aims to optimize heat pump performance, including under time-varying tariff conditions. The trial involved monitoring 76 properties with heat pumps, but without dedicated heat storage; 31 of these received the control system. Interviews were conducted with a subsample of 12 participants. The controller successfully evened out electricity demand over the day (reducing the evening peak), but this was associated with increased late night and daytime temperatures. Interview participants reported some disturbance owing to overnight heating and noise, as well as usability issues with the controller interface and hardware. These issues present risks to the future acceptability of such systems. While the system delivered short-term demand reductions successfully, longer-term demand shifting risked causing unacceptable disturbance to occupants. Future control systems could overcome some of the issues identified in this pioneering trial through more effective zoning, using temperature caps or installing dedicated heat storage, but these may either limit the available flexibility or be challenging to achieve.

Heat pumps (HPs) are seen as an increasingly important technology able to contribute significantly towards the decarbonisation of the domestic stock in the UK. However, there appears to be a performance gap between predicted and real-life HP performance, with several studies highlighting the need to include the HP’s interaction with users when examining their performance. This study examines the role of user behaviour in mitigating this performance gap from a systems perspective. A sample of 21 case studies was selected from 700 domestic HPs monitored across the UK via the government’s Renewable Heat Premium Payment Scheme for the collection of qualitative and quantitative socio-technical data. The application of systems thinking facilitated the identification of the underlying interactions between the HP system and its users. The systems analysis revealed that HP performance relies on complex socio-technical system interactions, including behavioural patterns, and that enabling feedback information processes can have a significant impact on user behaviour. The study enabled a deeper perspective on performance influencers relating to behavioural patterns and achieved new insights into the requirements for well-performing HPs. These findings have important implications for policy makers, installers and manufacturers of HP systems and their users.

Background: With the ongoing, rapid advances in Deep Learning (DL), such solutions can now detect medical conditions even invisible to the human eye. In this direction, efforts have been made to develop DL algorithms that diagnose paroxysmal atrial fibrillation (PAF) from electrocardiogram (ECG) signals in sinus rhythm (SR). However, many of the available approaches function as “black boxes”, with physicians unable to understand and trust their predictions. Purpose: To train a DL model to detect PAF patients while in SR and apply an algorithm that interprets and visualises its decisions. Methods: We obtained ECG samples from PAF and non-PAF patients during SR, from the PAF Prediction Challenge Database. After discarding unannotated samples and augmenting the sample size (by dividing each signal into 30-second segments), we split the whole dataset into a train (68%), a validation (16%) and a test (16%) set. No pair of samples belonging to different sets originated from the same patient. We trained the InceptionTime neural network on the train/validation sets and tested on the “unseen” test set after “hiding” the correct answers. Its performance was evaluated with the following metrics: Accuracy, f1-score, precision and recall (sensitivity). After repeating this process 20 times, we obtained a distribution for each score. Finally, we adjusted the Grad-CAM interpretation algorithm to our data and used it to visualise the areas perceived as important by the model. Results: After pre-processing, 4,080, 30-second, two-lead ECG signals were allocated to the train set, 960 to the validation and 960 to the test set. Each subset contained an equal number of PAF and non-PAF samples. After repeated training and testing, we obtained a median accuracy of 0.84 (interquartile range, IQR: 0.66–0.88), an f1-score of 0.82 (IQR: 0.68–0.88) and a median precision and recall equal to 0.93 (IQR: 0.67–0.99) and 0.77 (IQR: 0.68–0.93), respectively. The Grad-CAM technique highlighted the ECG areas of interest that led to each decision. We selected and present both PAF-positive and -negative samples, perceived either correctly or falsely. Interestingly, correct model decisions tend to focus on the P-wave, while false ones fixate on other regions. Conclusions: Although a pilot study with considerable limitations (small sample size, disregard of possible confounding due to comorbidities or other factors), this work shows how DL can be employed to distinguish between PAF and non-PAF patients from SR ECG samples, and confirms the potential of DL-enabled approaches to offer novel diagnostic capabilities. Most importantly, our effort provides a comprehensible, visual interpretation of the model's decisions. Demystifying DL behaviour can, not only improve such efforts by explaining false decisions, but also cultivate trust among clinicians and, possibly, point out directions for future research, since we can now see through the magnifying lens of a neural network. Funding Acknowledgement: Type of funding sources: None.