•Agonist-antagonist coactivation leads to unintended finger force increase.•This force increase is not perceived according to verbal reports on force magnitude.•Matching with contralateral hand ...reflects the increased force magnitude.•The findings fit the theory of movement control with spatial referent coordinates.
We tested several predictions of the theory of motor control with spatial referent coordinates related to effects of muscle coactivation on force production and perception. In particular, we predicted that subjects would produce unintentional force increase by finger flexors and be unaware of this force increase. Healthy subjects performed steady force production task in isometric conditions with visual feedback on the force level. They coactivated muscles of the arm trying to keep the force constant in the absence of visual feedback. This led to a consistent force increase not perceived by the subjects as reflected by their verbal reports. In contrast, when asked to match the force with the contralateral hand, adequate force matching was observed. Using the “inverse piano” apparatus confirmed no change in the referent coordinate of the fingers and an increase in its apparent stiffness. The results confirm the earlier hypothesis on the reciprocal command being hierarchically higher than the coactivation command. The observations suggest that verbal reports and force matching use different neural mechanisms of force perception; the former are dominated by sense of effort, which reflects primarily the magnitude of the reciprocal command. There were only minor differences between the dominant and non-dominant hands, likely reflecting the faster unintentional drifts of control variables in the dominant hand.
As the enthusiasm for integrating artificial intelligence (AI) into clinical care grows, so has our understanding of the challenges associated with deploying impactful and sustainable clinical AI ...models. Complex dataset shifts resulting from evolving clinical environments strain the longevity of AI models as predictive accuracy and associated utility deteriorate over time.
Responsible practice thus necessitates the lifecycle of AI models be extended to include ongoing monitoring and maintenance strategies within health system algorithmovigilance programs. We describe a framework encompassing a 360° continuum of preventive, preemptive, responsive, and reactive approaches to address model monitoring and maintenance from critically different angles.
We describe the complementary advantages and limitations of these four approaches and highlight the importance of such a coordinated strategy to help ensure the promise of clinical AI is not short-lived.
Machine learning has demonstrated success in clinical risk prediction modeling with complex electronic health record (EHR) data. However, the evolving nature of clinical practices can dynamically ...change the underlying data distribution over time, leading to model performance drift. Adopting an outdated model is potentially risky and may result in unintentional losses. In this paper, we propose a novel Hybrid Adaptive Boosting approach (HA‐Boost) for transfer learning. HA‐Boost is characterized by the domain similarity‐based and class imbalance‐based adaptation mechanisms, which simultaneously address two critical limitations of the classical TrAdaBoost algorithm. We validated HA‐Boost in predicting hospital‐acquired acute kidney injury using real‐world longitudinal EHRs data. The experiment results demonstrate that HA‐Boost stably outperforms the competing baselines in terms of both Area Under Receiver Operating Characteristic and Area Under Precision‐Recall Curve across a 7‐year time span. This study has confirmed the effectiveness of transfer learning as a superior model updating approach in a dynamic environment.
Background
Clinical prediction models suffer from performance drift as the patient population shifts over time. There is a great need for model updating approaches or modeling frameworks that can ...effectively use the old and new data.
Objective
Based on the paradigm of transfer learning, we aimed to develop a novel modeling framework that transfers old knowledge to the new environment for prediction tasks, and contributes to performance drift correction.
Methods
The proposed predictive modeling framework maintains a logistic regression–based stacking ensemble of 2 gradient boosting machine (GBM) models representing old and new knowledge learned from old and new data, respectively (referred to as transfer learning gradient boosting machine TransferGBM). The ensemble learning procedure can dynamically balance the old and new knowledge. Using 2010-2017 electronic health record data on a retrospective cohort of 141,696 patients, we validated TransferGBM for hospital-acquired acute kidney injury prediction.
Results
The baseline models (ie, transported models) that were trained on 2010 and 2011 data showed significant performance drift in the temporal validation with 2012-2017 data. Refitting these models using updated samples resulted in performance gains in nearly all cases. The proposed TransferGBM model succeeded in achieving uniformly better performance than the refitted models.
Conclusions
Under the scenario of population shift, incorporating new knowledge while preserving old knowledge is essential for maintaining stable performance. Transfer learning combined with stacking ensemble learning can help achieve a balance of old and new knowledge in a flexible and adaptive way, even in the case of insufficient new data.
We explored changes in the cyclical two-finger force performance task caused by turning visual feedback off performed either by the index and middle fingers of the dominant hand or by two index ...fingers of two persons. Based on an earlier study, we expected drifts in finger force amplitude and midpoint without a drift in relative phase. The subjects performed two rhythmical tasks at 1 Hz while paced by an auditory metronome. One of the tasks required cyclical changes in total force magnitude without changes in the sharing of the force between the two fingers. The other task required cyclical changes in the force sharing without changing total force magnitude. Subjects were provided with visual feedback, which showed total force magnitude and force sharing via cursor motion along the vertical and horizontal axes, respectively. Further, visual feedback was turned off, first on the variable that was not required to change and then on both variables. Turning visual feedback off led to a mean force drift toward lower magnitudes while force amplitude increased. There was a consistent drift in the relative phase in the one-hand task with the index finger leading the middle finger. No consistent relative phase drift was seen in the two-person tasks. The shape of the force cycle changed without visual feedback reflected in the lower similarity to a perfect cosine shape and in the higher time spent at lower force magnitudes. The data confirm findings of earlier studies regarding force amplitude and midpoint changes, but falsify predictions of an earlier proposed model with respect to the relative phase changes. We discuss factors that could contribute to the observed relative phase drift in the one-hand tasks including the leader–follower pattern generalized for two-effector tasks performed by one person.
This study investigates the impact of battery and fuel cell (FC) degradation on energy management of a FC hybrid electric vehicle. In this respect, an online energy management strategy (EMS) is ...proposed considering simultaneous online adaptation of battery and FC models. The EMS is based on quadratic programming which is integrated into an online battery and proton exchange membrane FC (PEMFC) parameters identification. Considering the battery and PEMFC states of health, three scenarios have been considered for the EMS purpose, and the performance of the proposed EMS has been examined under two driving cycles. Numerous test scenarios using standard driving cycles reveal that the ageing of battery and PEMFC has a considerable impact on the hydrogen consumption. Moreover, the proposed EMS can successfully tackle the model uncertainties owing to the performance drifts of the power sources at the mentioned scenarios.