Machine learning and geometric morphometrics to predict obstructive sleep apnea from 3D craniofacial scans.

BACKGROUND: Obstructive sleep apnea (OSA) remains massively underdiagnosed, due to limited access to polysomnography (PSG), the highly complex gold standard for diagnosis. Performance scores in predicting OSA are evaluated for machine learning (ML) analysis applied to 3D maxillofacial shapes. METHODS: The 3D maxillofacial shapes were scanned on 280 Caucasian men with suspected OSA. All participants underwent single night in-home or in-laboratory sleep testing with PSG (Nox A1, Resmed, Australia), with concomitant 3D scanning (Sense v2, 3D systems corporation, USA). Anthropometric data, comorbidities, medication, BERLIN, and NoSAS questionnaires were also collected at baseline. The PSG recordings were manually scored at the reference sleep center. The 3D craniofacial scans were processed by geometric morphometrics, and 13 different supervised algorithms, varying from simple to more advanced, were trained and tested. Results for OSAS recognition by ML models were then compared with scores for specificity and sensitivity obtained using BERLIN and NoSAS questionnaires. RESULTS: All valid scans (n = 267) were included in the analysis (patient mean age: 59 ± 9 years; BMI: 27 ± 4 kg/m2). For PSG-derived AHI≥15 events/h, the 56% specificity obtained for ML analysis of 3D craniofacial shapes was higher than for the questionnaires (Berlin: 50%; NoSAS: 40%). A sensitivity of 80% was obtained using ML analysis, compared to nearly 90% for NoSAS and 61% for the BERLIN questionnaire. The auROC score was further improved when 3D geometric morphometrics were combined with patient anthropometrics (auROC = 0.75). CONCLUSION: The combination of 3D geometric morphometrics with ML is proposed as a rapid, efficient, and inexpensive screening tool for OSA. TRIAL REGISTRATION NUMBER: NCT03632382; Date of registration: 15-08-2018.

Tracking atmospheric dispersion of metals in Rio de Janeiro Metropolitan region (Brazil) with epiphytes as bioindicators.

The atmospheric dispersion atmospheric plume of the metropolitan region and neighborhoods of the city of Rio de Janeiro was investigated through elemental analyzes (Na, K, Al, Mn, Pb, Ni, Cr) and stable lead isotopes in two epiphytes; the lichen Parmotrema crinitum (Ach.) and the bromeliad Tillandsia usneoides (L.). All the elements had lower concentrations in epiphytes than in rocks of the local geological basement, with the exception of K, which was similar to rocks. This behavior was attributed to the nutritional essentiality and abundance of this element in the environment. The concentrations of Na, Pb and the 206Pb/207Pb and 208Pb/206Pb ratios indicated the presence of spatial gradients in the path (a) sea shore → continent, and (b) urban center → farmlands. These patterns were associated with the mesoscale atmospheric circulation regime, in which, during the arrival of cold fronts, air masses from the metropolitan region of Rio de Janeiro are transported towards the mountainous region of Teresópolis, which acts as a barrier to the dispersion of pollutants. The bromeliad Tillandsia usneoides and the lichen Parmotrema crinitum were linked to Na and Pb atmospheric levels, respectively.

Mobility of Ni, Co, and Mn in ultramafic mining soils of New Caledonia, assessed by kinetic EDTA extractions.

The aim of this study was to determine the mobilization capability of Ni, Co, and Mn contained in New Caledonian ultramafic soils. Two series of soils were sampled: bare-surface mining soils in a Ni-mining context (n = 10), and forest soils, either in the vicinity of mine-working areas (n = 3) or far away from any known mining activity (n = 2). We focused on the < 100 µm soil fraction, because of its sensitivity to wind erosion, and its possible dissemination toward urbanized areas. In order to assess maximum potential metal mobility, EDTA kinetic extractions were performed over 24 h. Extraction curves were modeled as the sum of two first-order reactions. The first EDTA extracted pool corresponds to "quickly" released metals, while the second pool corresponds to "slowly" released metals. The remaining fraction is the EDTA non-extractable pool. Extractable Ni, Co, and Mn were always low in relation to total concentrations (< 5% for Ni, and 5-35% for Co and Mn). The extraction rate of the less labile pool was significantly higher for forest soils than for mining soils, whatever the metal. Despite the greater extractability potential in forest surface soils, mining soils represent a bigger environmental risk, because of their high metal content and, above all, because of their predisposition to surface runoff and eolian deflation.

Trace metals from historical mining sites and past metallurgical activity remain bioavailable to wildlife today.

Throughout history, ancient human societies exploited mineral resources all over the world, even in areas that are now protected and considered to be relatively pristine. Here, we show that past mining still has an impact on wildlife in some French protected areas. We measured cadmium, copper, lead, and zinc concentrations in topsoils and wood mouse kidneys from sites located in the Cévennes and the Morvan. The maximum levels of metals in these topsoils are one or two orders of magnitude greater than their commonly reported mean values in European topsoils. The transfer to biota was effective, as the lead concentration (and to a lesser extent, cadmium) in wood mouse kidneys increased with soil concentration, unlike copper and zinc, providing direct evidence that lead emitted in the environment several centuries ago is still bioavailable to free-ranging mammals. The negative correlation between kidney lead concentration and animal body condition suggests that historical mining activity may continue to play a role in the complex relationships between trace metal pollution and body indices. Ancient mining sites could therefore be used to assess the long-term fate of trace metals in soils and the subsequent risks to human health and the environment.

Impact of nickel mining in New Caledonia assessed by compositional data analysis of lichens.

The aim of this study is to explore the use of lichens as biomonitors of the impact of nickel mining and ore treatment on the atmosphere in the New Caledonian archipelago (South Pacific Ocean); both activities emitting also Co, Cr and possibly Fe. Metal contents were analysed in thirty-four epiphytic lichens, collected in the vicinity of the potential sources, and in places free from known historical mining. The highest Ni, Co, and Cr concentrations were, as expected, observed in lichens collected near ore deposits or treatment areas. The elemental composition in the lichens was explored by multivariate analysis, after appropriately transforming the variables (i.e. using compositional data analysis). The sample score of the first principal component (PC1) makes the largest (positive) multiplicative contribution to the log-ratios of metals originating from mining activities (Ni, Cr, Co) divided by Ti. The PC1 scores are used here as a surrogate of pollution levels related to mining and metallurgical activity. They can be viewed as synthetic indicators mapped to provide valuable information for the management and protection of ecosystems or, as a first step, to select locations where air filtration units could be installed, in the future, for air quality monitoring. However, as this approach drastically simplifies the problem, supplying a broadly efficient picture but little detail, recognizing the different sources of contamination may be difficult, more particularly when their chemical differences are subtle. It conveys only relative information: about ratios, not levels, and is therefore recommended as a preliminary step, in combination with close examination of raw concentration levels of lichens. Further validation using conventional air-monitoring by filter units should also prove beneficial.

Alternative dry separation of PM10 from soils for characterization by kinetic extraction: example of new Caledonian mining soils.

A simple new device for dry separation of fine particulate matter from bulk soil samples is presented here. It consists of a stainless steel tube along which a nitrogen flow is imposed, resulting in the displacement of particles. Taking into account particle transport, fluid mechanics, and soil sample composition, a tube 6-m long, with a 0.04-m diameter, was found best adapted for PM10 separation. The device rapidly produced several milligrams of particulate matter, on which chemical extractions with EDTA were subsequently performed to study the kinetic parameters of extractable metals. New Caledonian mining soils were chosen here, as a case-study. Although the easily extracted metal pool represents only 0.5-6.4 % of the total metal content for the elements studied (Ni, Co, Mn), the total concentrations are extremely high. This pool is therefore far from negligible, and can be troublesome in the environment. This dry technique for fine particle separation from bulk parent soil eliminates the metal-leaching risks inherent in wet filtration and should therefore ensure safe assessment of environmental quality in fine-textured, metal-contaminated soils.

Kinetic extractions to assess mobilization of Zn, Pb, Cu, and Cd in a metal-contaminated soil: EDTA vs. citrate.

Kinetic EDTA and citrate extractions were used to mimic metal mobilization in a soil contaminated by metallurgical fallout. Modeling of metal removal rates vs. time distinguished two metal pools: readily labile (QM1) and less labile (QM2). In citrate extractions, total extractability (QM1+QM2) of Zn and Cd was proportionally higher than for Pb and Cu. Proportions of Pb and Cu extracted with EDTA were three times higher than when using citrate. We observed similar QM1/QM2 ratios for Zn and Cu regardless of the extractant, suggesting comparable binding energies to soil constituents. However, for Pb and Cd, more heterogeneous binding energies were hypothesized to explain different kinetic extraction behaviors. Proportions of citrate-labile metals were found consistent with their short-term, in-situ mobility assessed in the studied soil, i.e., metal amount released in the soil solution or extracted by cultivated plants. Kinetic EDTA extractions were hypothesized to be more predictive for long-term metal migration with depth.

Apparent discrepancy in contamination history of a sub-tropical estuary evaluated through 210Pb profile and chronostratigraphical markers.

Zn and Cd concentrations, stable lead isotopes and 210Pb-derived chronology were determined in a sediment core sampled at Sepetiba Bay (South-eastern Brazil). During the last decades, the bay's watershed has been modified by the increase of industrial activities and human interventions. In particular, Zn and Cd ore treatment plants were built near the coast in 1960 and 1970, respectively, and water has been diverted from the adjacent Paraíba do Sul River watershed since 1950. The core collected at shallow depth near the industrial area exhibits four successive events: (i) at 50 cm depth, a change in the 206Pb/207Pb ratio from about 1.162 to more than 1.18 might be the result of the São Francisco Channel opening and water diversion from Paraíba do Sul river; (ii) at 40 cm depth, Zn concentration starts to increase (up to 0.8 mg g(-1)) (iii) above 30 cm depth, relatively high Cd concentrations (up to 1.6 microg g(-1)) are observed and (iv) at 16 cm depth, change in unsupported 210Pb slope is probably related to a waste dam built to prevent strong metal contamination in the bay. Sediment accumulation rates evaluated by Zn and Cd profiles used as time-markers are higher than those calculated from 210Pb-based chronology models. Using the constant initial concentration (CIC) model both events are supposed to date back to about 1884 and 1902, respectively, while using the constant rate of supply (CRS) model it shifts to about 1925 and 1935. Such discrepancies are probably assigned to the fact that these models do not take into account site-specific local sedimentation dynamics. In the study area, particles deposition seems to be controlled by enrichment with unsupported 210Pb transported by runoff from a mangrove flat bank. Chronology derived from a model that assumes an exponential increase of the initial 210Pb activity fits well with the estimated rates obtained from historical events.