Towards intelligent supervision of operating rooms using stencil-based character recognition.

The development of intelligent operating rooms is an example of a cyber-physical system resulting from the symbiosis of Industry 4.0 and medicine. A problem with this type of systems is that it requires demanding solutions that allow the real time acquisition of heterogeneous data in an efficient way. The aim of the presented work is the development of a data acquisition system, based on a real-time artificial vision algorithm which can capture information from different clinical monitors. The system was designed for the registration, pre-processing, and communication of clinical data recorded in an operating room. The methods for this proposal are based on a mobile device running a Unity application, which extracts information from clinical monitors and transmits the data to a supervision system through a wireless Bluetooth connection. The software implements a character detection algorithm and allows online correction of identified outliers. The results validate the system with real data obtained during surgical interventions, where only 0.42% values were missed and 0.89% were misread. The outlier detection algorithm was able to correct all the reading errors. In conclusion, the development of a low-cost compact solution to supervise operating rooms in real-time, collecting visual information non-intrusively and communicating data wirelessly, can be a very useful tool to overcome the lack of expensive data recording and processing technology in many clinical situations. The acquisition and pre-processing method presented in this article constitutes a key element towards the development of a cyber-physical system for the development of intelligent operating rooms.

Robust PID control of propofol anaesthesia: Uncertainty limits performance, not PID structure.

BACKGROUND AND OBJECTIVE: New proposals to improve the regulation of hypnosis in anaesthesia based on the development of advanced control structures emerge continuously. However, a fair study to analyse the real benefits of these structures compared to simpler clinically validated PID-based solutions has not been presented so far. The main objective of this work is to analyse the performance limitations associated with using a filtered PID controller, as compared to a high-order controller, represented through a Youla parameter. METHODS: The comparison consists of a two-steps methodology. First, two robust optimal filtered PID controllers, considering the effect of the inter-patient variability, are synthesised. A set of 47 validated paediatric pharmacological models, identified from clinical data, is used to this end. This model set provides representative inter-patient variability Second, individualised filtered PID and Youla controllers are synthesised for each model in the set. For fairness of comparison, the same performance objective is optimised for all designs, and the same robustness constraints are considered. Controller synthesis is performed utilising convex optimisation and gradient-based methods relying on algebraic differentiation. The worst-case performance over the patient model set is used for the comparison. RESULTS: Two robust filtered PID controllers for the entire model set, as well as individual-specific PID and Youla controllers, were optimised. All considered designs resulted in similar frequency response characteristics. The performance improvement associated with the Youla controllers was not significant compared to the individually tuned filtered PID controllers. The difference in performance between controllers synthesized for the model set and for individual models was significantly larger than the performance difference between the individual-specific PID and Youla controllers. The different controllers were evaluated in simulation. Although all of them showed clinically acceptable results, the robust solutions provided slower responses. CONCLUSION: Taking the same clinical and technical considerations into account for the optimisation of the different controllers, the design of individual-specific solutions resulted in only marginal differences in performance when comparing an optimal Youla parameter and its optimal filtered PID counterpart. The inter-patient variability is much more detrimental to performance than the limitations imposed by the simple structure of the filtered PID controller.

Machine learning based method for the evaluation of the Analgesia Nociception Index in the assessment of general anesthesia.

Measuring the level of analgesia to adapt the opioids infusion during anesthesia to the real needs of the patient is still a challenge. This is a consequence of the absence of a specific measure capable of quantifying the nociception level of the patients. Unlike existing proposals, this paper aims to evaluate the suitability of the Analgesia Nociception Index (ANI) as a guidance variable to replicate the decisions made by the experts when a modification of the opioid infusion rate is required. To this end, different machine learning classifiers were trained with several sets of clinical features. Data for training were captured from 17 patients undergoing cholecystectomy surgery. Satisfactory results were obtained when including information about minimum values of ANI for predicting a change of dose. Specifically, a higher efficiency of the Support Vector Machine (SVM) classifier was observed compared with the situation in which the ANI index was not included: accuracy: 86.21% (83.62%-87.93%), precision: 86.11% (83.78%-88.57%), recall: 91.18% (88.24%-91.18%), specificity: 79.17% (75%-83.33%), AUC: 0.89 (0.87-0.90) and kappa index: 0.71 (0.66-0.75). The results of this research evidenced that including information about the minimum values of ANI together with the hemodynamic information outperformed the decisions made regarding only non-specific traditional signs such as heart rate and blood pressure. In addition, the analysis of the results showed that including the ANI monitor in the decision making process may anticipate a dose change to prevent hemodynamic events. Finally, the SVM was able to perform accurate predictions when making different decisions commonly observed in the clinical practice.

Closed loop administration of propofol based on a Smith predictor: a randomized controlled trial.

BACKGROUND: Delay in the propofol pharmacodynamics effect is commonly observed in total intravenous anesthesia (TIVA). To face the delay in the hypnosis control, we have proposed a proportional-integral (PI) controller with a Smith predictor (PI+Smith). We have evaluated the feasibility of this closed-loop control for propofol administration and compared the performance with manual administration guided by the Bispectral Index (BIS). METHODS: Fifty-seven adult patients under TIVA with propofol and remifentanil were randomly assigned to a PI+Smith or a manual control (MC) group. The BIS target was set to 50. The performance was compared through the global score (GS), median performance error (MDPE), median absolute performance error (MDAPE), offset and Wobble. RESULTS: A total of 29 patients in the MC and 25 in the PI+Smith groups completed this study. Performance was significantly better in the PI+Smith group: global score was 25 (19 to 37) for PI+Smith versus 44 (32 to 57) for MC (P<0.001); MDPE was -0.9 (-5.6 to 2) for PI+Smith versus -11 (-16 to -4.3) for MC (P<0.001); MDAPE was 10.8 (8.8 to 14.3) for PI+Smith versus 17 (12.8 to 19.2) for MC (P<0.001); offset was -0.6 (-3.2 to 0.06) for PI+Smith versus -3.7 (-7.0 to -0.8) for MC (P=0.01). The percentage time of BIS within the 40-60 range during the maintenance phase was higher in the PI+Smith group 80.8 (68.7 to 87.9) than in the MC group 59.1 (53.4 to 72.5) (P<0.001). CONCLUSIONS: The use of a specific mechanism in the PI controller to deal with the delay outperformed satisfactorily manual practice. The controller was able to regulate propofol administration, maintaining the BIS value within a desirable range and coping with oscillations.

Improving the anesthetic process by a fuzzy rule based medical decision system.

OBJECTIVE: The main objective of this research is the design and implementation of a new fuzzy logic tool for automatic drug delivery in patients undergoing general anesthesia. The aim is to adjust the drug dose to the real patient needs using heuristic knowledge provided by clinicians. A two-level computer decision system is proposed. The idea is to release the clinician from routine tasks so that he can focus on other variables of the patient. METHODS: The controller uses the Bispectral Index (BIS) to assess the hypnotic state of the patient. Fuzzy controller was included in a closed-loop system to reach the BIS target and reject disturbances. BIS was measured using a BIS VISTA monitor, a device capable of calculating the hypnosis level of the patient through EEG information. An infusion pump with propofol 1% is used to supply the drug to the patient. The inputs to the fuzzy inference system are BIS error and BIS rate. The output is infusion rate increment. The mapping of the input information and the appropriate output is given by a rule-base based on knowledge of clinicians. RESULTS: To evaluate the performance of the fuzzy closed-loop system proposed, an observational study was carried out. Eighty one patients scheduled for ambulatory surgery were randomly distributed in 2 groups: one group using a fuzzy logic based closed-loop system (FCL) to automate the administration of propofol (42 cases); the second group using manual delivering of the drug (39 cases). In both groups, the BIS target was 50. CONCLUSIONS: The FCL, designed with intuitive logic rules based on the clinician experience, performed satisfactorily and outperformed the manual administration in patients in terms of accuracy through the maintenance stage.