Passive detection of COVID-19 with wearable sensors and explainable machine learning algorithms

Individual smartwatch or fitness band sensor data in the setting of COVID-19 has shown promise to identify symptomatic and pre-symptomatic infection or the need for hospitalization, correlations between peripheral temperature and self-reported fever, and an association between changes in heart-rate-variability and infection. In our study, a total of 38,911 individuals (61% female, 15% over 65) have been enrolled between March 25, 2020 and April 3, 2021, with 1118 reported testing positive and 7032 negative for COVID-19 by nasopharyngeal PCR swab test. We propose an explainable gradient boosting prediction model based on decision trees for the detection of COVID-19 infection that can adapt to the absence of self-reported symptoms and to the available sensor data, and that can explain the importance of each feature and the post-test-behavior for the individuals. We tested it in a cohort of symptomatic individuals who exhibited an AUC of 0.83 [0.81–0.85], or AUC = 0.78 [0.75–0.80] when considering only data before the test date, outperforming state-of-the-art algorithm in these conditions. The analysis of all individuals (including asymptomatic and pre-symptomatic) when self-reported symptoms were excluded provided an AUC of 0.78 [0.76–0.79], or AUC of 0.70 [0.69–0.72] when considering only data before the test date. Extending the use of predictive algorithms for detection of COVID-19 infection based only on passively monitored data from any device, we showed that it is possible to scale up this platform and apply the algorithm in other settings where self-reported symptoms can not be collected.

Passive Detection of COVID-19 with Wearable Sensors and Explainable Machine Learning Algorithms

Individual smartwatch or fitness band sensor data in the setting of COVID-19 has shown promise to identify symptomatic and pre-symptomatic infection or the need for hospitalization, correlations between peripheral temperature and self-reported fever, and an association between changes in heart-rate-variability and infection. In our study, a total of 38,911 individuals (61% female, 15% over 65) have been enrolled between March 25, 2020 and April 3, 2021, with 1,118 reported testing positive and 7,032 negative for COVID-19 by nasopharyngeal PCR swab test. We propose an explainable gradient boosting prediction model based on decision trees for the detection of COVID-19 infection that can adapt to the absence of self-reported symptoms and to the available sensor data, and that can explain the importance of each feature and the post-test-behavior for the individuals. We tested it in a cohort of symptomatic individuals who exhibited an AUC of 0.83 [0.81-0.85], or AUC=0.78 [0.75-0.80] when considering only data before the test date, outperforming state-of-the-art algorithm in these conditions. The analysis of all individuals (including asymptomatic and pre-symptomatic) when self-reported symptoms were excluded provided an AUC of 0.78 [0.76-0.79], or AUC of 0.70 [0.69-0.72] when considering only data before the test date. Extending the use of predictive algorithms for detection of COVID-19 infection based only on passively monitored data from any device, we showed that it is possible to scale up this platform and apply the algorithm in other settings where self-reported symptoms can not be collected.

Spatial neglect in the digital age – Influence of presentation format on patients' test behavior

Computerization of diagnostic neglect tests can deepen our knowledge of neglect specific abnormalities, by effortlessly providing additional behavioral markers that are hardly extractable from existing paper-and-pencil versions. However, so far it is not known whether the digitization and/or a change in size format impact neglect patients' search behavior and test scores and thus require adjustments of cut-off criteria. We compared the Center of cancellation (CoC) measure of right hemisphere stroke patients with spatial neglect in two cancellation tasks across different modalities (paper-and-pencil vs. digital) and display sizes (small, medium, large). We found that the CoC measure did neither vary considerably between paper-and-pencil versus digital versions, nor between different digital size formats. The CoC derived from cancellation tasks thus seems robust to test digitization. A further aim of the present study was to evaluate three additional parameters of search behavior which became available through digitization. We observed slower search behavior, increased distance between two consecutively identified items, and signs of a more strategic search for neglect patients than control patients without neglect. Machine learning classifications indicated that - beyond the CoC measure - the latter three variables can help to differentiate stroke patients with spatial neglect from those without.

Continuous automated monitoring of the homecage behavior of mice: a small world analysis

The automation of monitoring and analysis of mouse behaviour in a homecage can be obtained from continuous video records with machine learning and computer vision. The approach of recreating a mouse’s “real world” behavior and laboratory test behavior in the “small world” of a laboratory cage can provide insights into phenotypical expression of mouse genotypes, development and aging, and neurological disease. Algorithms identify behavioral acts (walk, rear), actions (sleep duration, distance travelled), organized patterns of movement (home base activity and excursions) over extended periods of time. In addition, performance on specific tests can be incorporated within a mouse’s living arrangement. Here we review approaches to engineering a small world and state of the art machine learning analyses for automated study of mouse homecage behavior. We highlight advantages and limitations of these approaches as a supplement to acute behavioral testing methodology.

GRID LASER MARKING INFLUENCE ON HIGH-STRENGTH STEELS TENSILE TEST BEHAVIOR

Automotive car companies are using AHSS (advanced high strength steels) over the last 20 years, to reduce vehicle weight and improve safety. The new steels can achieve higher strength and good fatigue resistance, but some issues related to springback and low formability are also a big concern. Thus, companies need to extend their know-how regarding material behaviour, design rules and manufacturing processes. Therefore, materials characterization laboratories are working to obtain the new formability charts of the steels. The grid laser marking of test pieces is a recent approach. However, the marking process must accomplish three main aspects: indelibility during the tensile testing procedure, precision, and of course, it must not affect the mechanical properties of studied steels. This work is focused on the laser marking of test pieces, using Ytterbium fiber laser. A dual phase steel (JFE CA 1180) is studied. Process parameter are defined. Keywords: grid marking, laser, advanced high-strength steels, AHSS, formability diagrams, mechanical properties

Health Education (HE) Effectiveness Towards Knowledge and Behavior of the Treatment of Kanguru Method (PMK) in Public Women in Perinatology Room Luwuk Hospital, Banggai District, Central Sulawesi Province

Background: Up to now, the number of preterm births in Indonesia has reached 10-15% of total births each year. The results of a preliminary study conducted by researchers on November 11-16 2019 in the Perinatology Room at Luwuk Regional Hospital, Banggai District, Central Sulawesi Province, with interviews with 10 infant mothers, it was found that 5 babies said they did not know about treatment using the kangaroo method. The statement of one baby mother said that the mother's knowledge was still low about FMD because the kangaroo method was still considered by the mothers. The research objective was to determine the effectiveness of Health Education (HE) on knowledge about Kangaroo (PMK) treatment of postpartum mothers in the perinatology room of Luwuk Regional Hospital, Banggai Regency, Central Sulawesi Province. Methods: This study used a one group pre-test-post-test design. The number of samples is 35 respondents. Sampling using Accidental Sampling technique. Statistical test using Paired Samples T-Test. The results showed that the pre-test score of knowledge obtained an average of 45.57 and the post-test score of knowledge obtained an average of 74.43. Meanwhile, the pre-test behavior score obtained an average of 43.10 and the post-test behavior score obtained an average of 67.62. The results of the Paired Samples T-Test for knowledge showed a significance value of 0.000 <0.05 and behavior known a significance value of 0.000 <0.05. Conclusion: that there is an average difference between the Pre Test and Post Test values, which means that Health Education (HE) can be said to be effective in increasing knowledge and behavior about Kangaroo Method Care (PMK) for postpartum mothers in the Perinatology Room of Luwuk Regional Hospital, Banggai Regency, Central Sulawesi Province .

The Evolving Basis for the Design of Light Gauge Steel Systems

The importance of allowing for the many different types of structural interaction that have an effect on the performance of light gauge members when used in practical situations is emphasized. A distinction is drawn between internal interactions involving the various plate elements of the steel profiles and external interactions involving the other components in the system. Although full-scale testing of representative systems can capture this behavior, the costs involved make this an impractical general basis for design; codified methods generally consider only isolated plates within members and isolated members within systems, thereby neglecting the potentially beneficial effects of both forms of interaction. Properly used, modern methods of numerical analysis offer the potential to systematically allow for both forms of interaction — provided the numerical models used have been adequately validated against suitable tests. The use of such an approach is explained and illustrated for three commonly used structural systems: roof purlins, floor beams, and columns in stud walls. In each case, it is shown that, provided sufficient care is taken, the numerical approach can yield accurate predictions of the observed test behavior. The subsequently generated large portfolio of numerical results can then provide clear insights into the exact nature of the various interactions and, thus, form the basis for more realistic design approaches that are both more accurate in their predictions and which lead to more economic designs. Building on this, modifying existing arrangements so as to yield superior performance through specific modifications is now possible. Two such examples, one in which improved interconnection between the components in a system is investigated and a second in which prestressing is shown to provide substantial enhancement for relatively small and simple changes, are presented.