Probabilistic graphical modelling of early childhood caries development.

In order to develop a statistical model for complex interactions among factors affecting early childhood caries development (ECC), 234 children from the "XXXX oral growth longitudinal study" were analysed at ages 3, 4 and 5. A questionnaire for children's parents (clinical history, nutritional and oral hygiene habits) and clinical oral examinations were recorded each year. The probabilistic dependence structure on the most significant factors was modelled with an Undirected Graphical Model (UGM or Markov random fields) which provides a probabilistic reasoning approach for the establishment of multi-way associations. The best fitting UGM was estimated through the maximum likelihood principle and two-way factor associations were verified through Fisher's exact statistical hypothesis tests for count data. The effect of sugar intake on caries incidence is mediated by oral hygiene; for caries incidence, oral hygiene quality is more relevant than toothbrushing time; the effect of pacifier on caries incidence is statistically significant only when considered in combination with breastfeeding time. Among behavioural ECC risk factors, the quality of oral hygiene, and not the toothbrushing frequency, is a primary factor that modulates the sugar intake in his primary role of the ECC developer. School-based oral health programs for ECC prevention should be improved with supervised toothbrushing program. UGM analysis could improve the school-based oral health programs with more effective and efficient prevention strategies based on the hierarchical interactions among the ECC risk factors. Oral hygiene plays a pivotal role in early childhood caries and can modulate positively or negatively their development. Supervised toothbrushing is a crucial intervention to be included in the daily educational and clinical practice and in the school-based oral health promotion programs. Trial registration: Clinical trial registration number: NCT02798809.

Self-Sufficient Sensor Node Embedding 2D Visible Light Positioning through a Solar Cell Module.

Nowadays, indoor positioning (IP) is a relevant aspect in several scenarios within the Internet of Things (IoT) framework, e.g., Industry 4.0, Smart City and Smart Factory, in order to track, amongst others, the position of vehicles, people or goods. This paper presents the realization and testing of a low power sensor node equipped with long range wide area network (LoRaWAN) connectivity and providing 2D Visible Light Positioning (VLP) features. Three modulated LED (light emitting diodes) sources, the same as the ones commonly employed in indoor environments, are used. The localization feature is attained from the received light intensities performing optical channel estimation and lateration directly on the target to be localized, equipped with a low-power microcontroller. Moreover, the node exploits a solar cell, both as optical receiver and energy harvester, provisioning energy from the artificial lights used for positioning, thus realizing an innovative solution for self-sufficient indoor localization. The tests performed in a ~1 m2 area reveal accurate positioning results with error lower than 5 cm and energy self-sufficiency even in case of radio transmissions every 10 min, which are compliant with quasi-real time monitoring tasks.