Innovative Machine Learning Algorithm Research Articles

Efficient machine learning for strength prediction of ready-mix concrete production (prolonged mixing)

Purpose This paper aims to clarify the efficient process of the machine learning algorithms implemented in the ready-mix concrete (RMC) onsite. It proposes innovative machine learning algorithms in terms of preciseness and computation time for the RMC strength prediction. Design/methodology/approach This paper presents an investigation of five different machine learning algorithms, namely, multilinear regression, support vector regression, k-nearest neighbors, extreme gradient boosting (XGBOOST) and deep neural network (DNN), that can be used to predict the 28- and 56-day compressive strengths of nine mix designs and four mixing conditions. Two algorithms were designated for fitting the actual and predicted 28- and 56-day compressive strength data. Moreover, the 28-day compressive strength data were implemented to predict 56-day compressive strength. Findings The efficacy of the compressive strength data was predicted by DNN and XGBOOST algorithms. The computation time of the XGBOOST algorithm was apparently faster than the DNN, offering it to be the most suitable strength prediction tool for RMC. Research limitations/implications Since none has been practically adopted the machine learning for strength prediction for RMC, the scope of this work focuses on the commercially available algorithms. The adoption of the modified methods to fit with the RMC data should be determined thereafter. Practical implications The selected algorithms offer efficient prediction for promoting sustainability to the RMC industries. The standard adopting such algorithms can be established, excluding the traditional labor testing. The manufacturers can implement research to introduce machine learning in the quality controcl process of their plants. Originality/value Regarding literature review, machine learning has been assessed regarding the laboratory concrete mix design and concrete performance. A study conducted based on the on-site production and prolonged mixing parameters is lacking.

Efficient machine learning for strength prediction of ready-mix concrete production (prolonged mixing)

Purpose This paper aims to clarify the efficient process of the machine learning algorithms implemented in the ready-mix concrete (RMC) onsite. It proposes innovative machine learning algorithms in terms of preciseness and computation time for the RMC strength prediction. Design/methodology/approach This paper presents an investigation of five different machine learning algorithms, namely, multilinear regression, support vector regression, k-nearest neighbors, extreme gradient boosting (XGBOOST) and deep neural network (DNN), that can be used to predict the 28- and 56-day compressive strengths of nine mix designs and four mixing conditions. Two algorithms were designated for fitting the actual and predicted 28- and 56-day compressive strength data. Moreover, the 28-day compressive strength data were implemented to predict 56-day compressive strength. Findings The efficacy of the compressive strength data was predicted by DNN and XGBOOST algorithms. The computation time of the XGBOOST algorithm was apparently faster than the DNN, offering it to be the most suitable strength prediction tool for RMC. Research limitations/implications Since none has been practically adopted the machine learning for strength prediction for RMC, the scope of this work focuses on the commercially available algorithms. The adoption of the modified methods to fit with the RMC data should be determined thereafter. Practical implications The selected algorithms offer efficient prediction for promoting sustainability to the RMC industries. The standard adopting such algorithms can be established, excluding the traditional labor testing. The manufacturers can implement research to introduce machine learning in the quality controcl process of their plants. Originality/value Regarding literature review, machine learning has been assessed regarding the laboratory concrete mix design and concrete performance. A study conducted based on the on-site production and prolonged mixing parameters is lacking.

Application of innovative SVM-PSO-GA algorithm to study vibrations of improved perovskite solar cells

This study investigates the vibrations of graphene oxide powders (GOPs) reinforced perovskite solar cells surrounded by an elastic foundation using both mathematical modeling and innovative machine learning algorithms. The incorporation of GOPs into the perovskite matrix enhances the mechanical properties and stability of the solar cells, which are crucial for their durability and efficiency. The analysis is conducted through the application of Hamilton’s principle, providing a robust theoretical framework for deriving the governing equations of motion. An analytical method is employed to solve these equations, allowing for the accurate prediction of the vibrational behavior of the reinforced solar cells. The effects of various parameters, including the stiffness of the elastic foundation and the concentration of GOPs, are systematically examined. This study presents the application of an innovative Support Vector Machine (SVM)-Particle Swarm Optimization (PSO)-Genetic Algorithm (GA) to analyze the vibrations of GOPs reinforced perovskite solar cells surrounded by an elastic foundation using mathematical modeling datasets. The SVM-PSO-GA algorithm to enhance predictive accuracy. The integrated approach leverages the strengths of each method to model and predict the vibrational behavior of the reinforced solar cells. The results highlight the algorithm’s effectiveness in capturing complex interactions and optimizing design parameters, providing valuable insights for improving the stability and performance of perovskite solar cells in practical applications.

Prediction of Student Performance Using Random Forest Combined With Naïve Bayes

Abstract Random forest is a powerful ensemble learning technique celebrated for its heightened predictive performance and robustness in handling complex datasets; nevertheless, it is criticized for its computational expense, particularly with a large number of trees in the ensemble. Moreover, the model’s interpretability diminishes as the ensemble’s complexity increases, presenting challenges in understanding the decision-making process. Although various pruning techniques have been proposed by researchers to tackle these issues, achieving a consensus on the optimal strategy across diverse datasets remains elusive. In response to these challenges, this paper introduces an innovative machine learning algorithm that integrates random forest with Naïve Bayes to predict student performance. The proposed method employs the Naïve Bayes formula to evaluate random forest branches, classifying data by prioritizing branches based on importance and assigning each example to a single branch for classification. The algorithm is utilized on two sets of student data and is evaluated against seven alternative machine-learning algorithms. The results confirm its strong performance, characterized by a minimal number of branches.

Innovative Machine Learning Algorithms for Classification and Intrusion Detectionv By IJISRT

With the escalating sophistication of cyber threats, the need for robust intrusion detection systems has become paramount in safeguarding information systems. This research addresses the limitations of traditional methods by proposing and evaluating innovative machine learning algorithms for classification in intrusion detection. The study explores a diverse set of algorithms designed to enhance accuracy, efficiency, and adaptability in the dynamic landscape of cybersecurity. The introduction provides a context for the research, emphasizing the critical role of intrusion detection in contemporary cybersecurity. A comprehensive literature review underscores the shortcomings of existing methodologies and sets the stage for the introduction of novel machine learning approaches. The research methodology outlines the dataset, evaluation metrics, and the training/testing process, ensuring transparency and replicability. The heart of the paper lies in the exploration of innovative machine learning algorithms. Each algorithm is introduced, highlighting unique features and innovations. The experimental results showcase the performance of these algorithms, with detailed comparisons against traditional counterparts. The discussion section interprets the results, emphasizing the practical implications and potential advancements these algorithms bring to the field. Addressing challenges encountered during implementation, the paper outlines future directions for research, providing a roadmap for continued innovation. The conclusion succinctly summarizes key findings, accentuating the groundbreaking contributions of the proposed machine learning algorithms to intrusion detection. This research significantly advances the discourse on intrusion detection systems, offering a paradigm shift towards more effective and adaptive solutions in the face of evolving cyber threats.

Automated Detection and Characterization of Wave Structures Obtained From GNSS Measurements

AbstractA simple technique designed to automatically identify and characterize wave structures from total electron content (TEC) data obtained from Global Navigation Satellite System (GNSS) satellites and multiple receiver stations is presented. We used 11 years of GNSS data (one complete solar cycle) to detect and characterize the traveling ionospheric disturbances (TIDs) (or wave structures) over high latitude (65°N ± 5°, 147°W ± 5°—Poker Flat, AK) and middle latitude (40°N ± 5°, 117°W ± 5°—Mt. Moses, NV) regions. The algorithm is capable of automatically detecting basic wave parameters such as wave period, horizontal phase velocity, amplitude, wave propagation direction, and wavelength. The designed algorithm can be applicable in the following areas: (a) global GNSS‐TEC (b) climatology of waves (c) input into innovative machine learning algorithms, such as ABCGAN (e.g., Valentic, 2023, https://doi.org/10.5281/zenodo.7747377 designed to characterize background ionospheric plasma and predict wave‐like high/low‐frequency perturbations) (d) anomaly detection for real‐time scenarios. Furthermore, we apply the developed wave detect and characterization technique to investigate three rockets launched on 26 and 28 January 2015. We examine the distribution of different scales of TIDs, and how they varied from high (Poker Flat) to middle (Mt. Moses) latitudes. Lastly, we show that the wave structures at the high‐latitude regions of Poker Flat are substantially affected by auroral processes and those from the middle‐latitude regions of Mt. Moses are impacted by AGWs coupling from below.

Dynamic responses of functionally graded origami-enabled auxetic metamaterial sector plate induced by mechanical shock: Application of innovative machine learning algorithm

For the first time, in the current work, the quasi-3D new refined theory (Q3D-NRT) is used to analyze the dynamic reactions of functionally graded (FG) annular sector plates made of graphene origami (GOri)-enabled auxetic metal metamaterials to the mechanical shock loading. The auxetic property of the annular plates is effectively controlled by graphene content and GOri folding degree that is graded across the thickness direction of the annular plates in a layer-wise manner, as Poisson’s ratio and other material properties are position-dependent and can be estimated by genetic programming-assisted micromechanical models (MM). Hamilton’s concept is used to find the structure’s governing equations. The differential quadrature technique and the Laplace transform are used at design sites to solve the time-varying equations of motion. The system response is translated from the Laplace domain to the time domain using a modified version of the Dubner and Abate approach. To investigate the impact of different geometrical and physical factors on the dynamic reactions of the annular sector plates, thorough parametric research is conducted. The findings of the present mathematical modeling are compared with those of the earlier publications and also with those of the machine learning approach. Utilizing this learning strategy, differential equations may be solved with very cheap computer costs while also overcoming formulation complexity. It needs a valid dataset from experimental or numerical analysis to use machine learning techniques. This dataset was compiled using the study’s numerical findings. Additionally, the machine learning approach demonstrates the capacity to deliver findings with a high degree of accuracy when predicting the mechanical characteristics of the existing structure under novel loading and boundary circumstances.

Improving the efficiency and stability of perovskite solar cell: Application of innovative machine learning algorithm

Solar cells are modern inventions that use the photovoltaic effect to directly convert light energy into electricity, generating electrical charges that are free to move through semiconductors. The semiconductor is typically utilized as the raw material for solar cells. In order to convert energy, electron–hole pairs that are responsible for producing light (photon) energy must be absorbed in a semiconductor, followed by charge carrier separation. Enhancing solar cells’ stability is crucial in engineering because they are used in a variety of environments. Moreover, graphene nanoplatelets (GPLs) have a great deal of potential to enhance ceramic–GNP composites’ mechanical, tribological, electrical, thermal, and biological characteristics, all at once. Machine learning algorithms (MLA) are often used to forecast how various systems would behave. In an MLA network, hyperparameters like the number of hidden layers and learning rate are often selected manually as required. MLA is used in this work to examine spinning cylindrical constructions’ stability at the microscale. In this context, the particle swarm optimization (PSO) is used to optimize the weights and biases of the network. The number of perceptions in the two hidden layers is optimized in a second parallel process using a genetic algorithm. The modified torque–stress theory (MCST) equation’s numerical solution for the dynamic behavior of GPL-reinforced perovskite solar cells was used to train the MLA. It is proposed to guide the spatial discretization of governing equations using the variational differential quadrature (VDQ) method as a direct discretization of the energy functional in the space domain. Lastly, the findings demonstrate that the stability of the current cantilevered solar cell reinforced by GPLs is significantly influenced by curvature, length scale, the shape of the solar cell, and mode number factors.

Identification of different MRI atrophy progression trajectories in epilepsy by subtype and stage inference.

Artificial intelligence (AI)-based tools are widely employed, but their use for diagnosis and prognosis of neurological disorders is still evolving. Here we analyse a cross-sectional multicentre structural MRI dataset of 696 people with epilepsy and 118 control subjects. We use an innovative machine-learning algorithm, Subtype and Stage Inference, to develop a novel data-driven disease taxonomy, whereby epilepsy subtypes correspond to distinct patterns of spatiotemporal progression of brain atrophy.In a discovery cohort of 814 individuals, we identify two subtypes common to focal and idiopathic generalized epilepsies, characterized by progression of grey matter atrophy driven by the cortex or the basal ganglia. A third subtype, only detected in focal epilepsies, was characterized by hippocampal atrophy. We corroborate external validity via an independent cohort of 254 people and confirm that the basal ganglia subtype is associated with the most severe epilepsy.Our findings suggest fundamental processes underlying the progression of epilepsy-related brain atrophy. We deliver a novel MRI- and AI-guided epilepsy taxonomy, which could be used for individualized prognostics and targeted therapeutics.

Yapay Sinir Ağları Kullanılarak Pediatrik Akciğer Röntgen Görüntülerinden Pnömoni Tespiti

Pneumonia Detection from Pediatric Lung X-Ray Images Using Artificial Neural Networks
 ABSTRACT
 Studies on medical imaging have grown significantly in recent years. Doctors have a crucial convenience for diagnosis thanks to semi- or fully automatic region recognition in medical imaging. It is crucial to support treatment without a specialist doctor, particularly in those nations where there is a dearth of such medical professionals. The little air sacs known as alveoli are most impacted by pneumonia, a lung inflammation. A key component of providing the right therapy conditions to heal patients and reduce harm while eradicating inflammation is early detection and precise diagnosis. Noise and blurring in patient photos obtained from X-ray machines are cleaned using deep learning algorithms and image processing techniques, and they are very helpful in. In this study, we studied chest X-ray images of pediatric patients with pneumonia and healthy individuals. XGBoost (eXtreme gradient boosting) is an innovative machine learning algorithm based on decision tree and using gradient boosting in its computations. It achieved 97.01% success with high classification performance.
 Keywords: Medical imaging, Machine learning, Pediatric Chest X-ray

Predicting the Gas Permeability of Sustainable Cement Mortar Containing Internal Cracks by Combining Physical Experiments and Hybrid Ensemble Artificial Intelligence Algorithms.

The presence of internal fissures holds immense sway over the gas permeability of sustainable cement mortar, which in turn dictates the longevity and steadfastness of associated edifices. Nevertheless, predicting the gas permeability of sustainable cement mortar that contains internal cracks poses a significant challenge due to the presence of numerous influential variables and intricate interdependent mechanisms. To solve the deficiency, this research establishes an innovative machine learning algorithm via the integration of the Mind Evolutionary Algorithm (MEA) with the Adaptive Boosting Algorithm-Back Propagation Artificial Neural Network (ABA-BPANN) ensemble algorithm to predict the gas permeability of sustainable cement mortar that contains internal cracks, based on the results of 1452 gas permeability tests. Firstly, the present study employs the MEA-tuned ABA-BPANN model as the primary tool for gas permeability prediction in cement mortar, a comparative analysis is conducted with conventional machine learning models such as Particle Swarm Optimisation Algorithm (PSO) and Genetic Algorithm (GA) optimised ABA-BPANN, MEA optimised Extreme Learning Machine (ELM), and BPANN. The efficacy of the MEA-tuned ABA-BPANN model is verified, thereby demonstrating its proficiency. In addition, the sensitivity analysis conducted with the aid of the innovative model has revealed that the gas permeability of durable cement mortar incorporating internal cracks is more profoundly affected by the dimensions and quantities of such cracks than by the stress conditions to which the mortar is subjected. Thirdly, puts forth a novel machine-learning model, which enables the establishment of an analytical formula for the precise prediction of gas permeability. This formula can be employed by individuals who lack familiarity with machine learning skills. The proposed model, namely the MEA-optimised ABA-BPANN algorithm, exhibits significant potential in accurately estimating the gas permeability of sustainable cement mortar that contains internal cracks in varying stress environments. The study highlights the algorithm's ability to offer essential insights for designing related structures.

A Deep Learning-Driven Self-Conscious Distributed Cyber-Physical System for Renewable Energy Communities.

The Internet of Things (IoT) is transforming various domains, including smart energy management, by enabling the integration of complex digital and physical components in distributed cyber-physical systems (DCPSs). The design of DCPSs has so far been focused on performance-related, non-functional requirements. However, with the growing power consumption and computation expenses, sustainability is becoming an important aspect to consider. This has led to the concept of energy-aware DCPSs, which integrate conventional non-functional requirements with additional attributes for sustainability, such as energy consumption. This research activity aimed to investigate and develop energy-aware architectural models and edge/cloud computing technologies to design next-generation, AI-enabled (and, specifically, deep-learning-enhanced), self-conscious IoT-extended DCPSs. Our key contributions include energy-aware edge-to-cloud architectural models and technologies, the orchestration of a (possibly federated) edge-to-cloud infrastructure, abstractions and unified models for distributed heterogeneous virtualized resources, innovative machine learning algorithms for the dynamic reallocation and reconfiguration of energy resources, and the management of energy communities. The proposed solution was validated through case studies on optimizing renewable energy communities (RECs), or energy-aware DCPSs, which are particularly challenging due to their unique requirements and constraints; in more detail, in this work, we aim to define the optimal implementation of an energy-aware DCPS. Moreover, smart grids play a crucial role in developing energy-aware DCPSs, providing a flexible and efficient power system integrating renewable energy sources, microgrids, and other distributed energy resources. The proposed energy-aware DCPSs contribute to the development of smart grids by providing a sustainable, self-consistent, and efficient way to manage energy distribution and consumption. The performance demonstrates our approach's effectiveness for consumption and production (based on RMSE and MAE metrics). Our research supports the transition towards a more sustainable future, where communities adopting REC principles become key players in the energy landscape.

Visual MMC Open Circuit Fault Real-Time Rapid Detection System

With the widespread use of modular multi-stage converters, the demands on their stability are increasing. In particular, problems such as open-circuit and short-circuit faults in their submodules have also attracted considerable attention from all walks of life. On this basis, a machine learning-based fault self-test and sub-module tracking strategy as well as innovative machine learning algorithms are proposed. Starting from the output characteristics of the sub-module, the harmonic components are analysed, the eigenvalues of the current system sub-module during normal operation and during faults are extracted, the eigenvalues are quickly categorised, and after categorisation, a new support vector machine model is put into place for machine learning. The trained machine model is finally embedded on top of the MCU integrated system and a communication transmission module is added on top of it, which can quickly determine the fault item in time when the system is running fault and reduce the maintenance cost later.

A Prototype Machine Learning Tool Aiming to Support 3D Crowdsourced Cadastral Surveying of Self-Made Cities

Land administration and management systems (LAMSs) have already made progress in the field of 3D Cadastre and the visualization of complex urban properties to support property markets and provide geospatial information for the sustainable management of smart cities. However, in less developed economies, with informally developed urban areas—the so-called self-made cities—the 2D LAMSs are left behind. Usually, they are less effective and mainly incomplete since a large number of informal constructions remain unregistered. This paper presents the latest results of an innovative on-going research aiming to structure, test and propose a low-cost but reliable enough methodology to support the simultaneous and fast implementation of both 2D land parcel and 3D property unit registration of informal, multi-story and unregistered constructions. An Indoor Positioning System (IPS) built upon low-cost Bluetooth technology combined with an innovative machine learning algorithm and connected with a 3D LADM-based cadastral mapping mobile application are the two key components of the technical solution under investigation. The proposed solution is tested for the first floor of a multi-room office building. The main conclusions concern the potential, usability and reliability of the method.

Accuracy of machine learning algorithms for the assessment of upper-limb motor impairments in patients with post-stroke hemiparesis: A systematic review and meta-analysis.

The assessment of motor function is vital in post-stroke rehabilitation protocols, and it is imperative to obtain an objective and quantitative measurement of motor function. There are some innovative machine learning algorithms that can be applied in order to automate the assessment of upper extremity motor function. To perform a systematic review and meta-analysis of the efficacy of machine learning algorithms for assessing upper limb motor function in post-stroke patients and compare these algorithms to clinical assessment. The protocol was registered in the International Prospective Register of Systematic Reviews (PROSPERO) database. The review was carried out according to the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) guidelines and the Cochrane Handbook for Systematic Reviews of Interventions. The search was performed using 6 electronic databases. The meta-analysis was performed with the data from the correlation coefficients using a random model. The initial search yielded 1626 records, but only 8 studies fully met the eligibility criteria. The studies reported strong and very strong correlations between the algorithms tested and clinical assessment. The meta-analysis revealed a lack of homogeneity (I2 = 85.29%, Q = 48.15), which is attributable to the heterogeneity of the included studies. Automated systems using machine learning algorithms could support therapists in assessing upper extremity motor function in post-stroke patients. However, to draw more robust conclusions, methodological designs that minimize the risk of bias and increase the quality of the methodology of future studies are required.

A study on quality control using delta data with machine learning technique

BackgroundIn the big data era, patient-based real-time quality control (PBRTQC), as an emerging quality control (QC) method, is expanding within the clinical laboratory industry. However, the main issue of current PBRTQC methodology is data stability. Our study is aimed to explore a novel protocol for data stability by combining delta data with machine learning (ML) technique to improve the capacity of QC event detection. MethodsA data set of 423,290 laboratory results from Beijing Chao-yang Hospital 2019 patient results were used as a training set (n = 380960, 90%) and internal validation set (n = 42330, 10%). A further 22,460 results from Beijing Long-fu Hospital 2019 patient results were used as a test set. Three-type data (1) Single-type data processed by truncation limits; (2) delta-type data processed by truncation limits and (3)delta-type data processed by Isolated Forest (IF) algorithm were evaluated with accuracy, sensitivity, NPed, etc., and compared with previously published statistical methods. ResultsThe optimal model was based on Random Forest (RF) algorithm by using delta-type data processed by IF algorithm. The model had a better accuracy (0.99), sensitivity (0.99) specificity (0.99) and AUC (0.99) with the dependent test set, surpassing the critical bias of PBRTQC by over 50%. For the LYMPH#, HGB, and PLT, the cumulative MNPed of MLQC were reduced by 95.43%, 97.39%, and 97.97% respectively when compared to the best of the PBRTQC. ConclusionFinal results indicate that by integrating an innovative ML algorithm with the overall data processing protocol the detection of QC events is improved.

Optimization of Thermal Modeling Using Machine Learning Techniques in Fused Deposition Modeling 3-D Printing

Abstract In this study, the cooler type produced with a fused deposition modeling (FDM) 3-D printer, one of the 3-D printing technologies, was investigated using image processing techniques and machine learning algorithms. This study aims to change the cooler design concept used in FDM 3-D printers and use image processing techniques and innovative machine learning algorithms to solve the temperature effect problems on the part. In this study, four different cooler types— no-cooler, A-type, B-type, and C-type—were used with an FDM 3-D printer, and each layer processing image of these parts was captured with a thermal camera. Temperature distribution diagrams of the parts were drawn according to layers using image processing techniques such as the Gaussian filtering method and the Sobel and Canny edge detection techniques. Using three different machine learning algorithms on the temperature data set obtained from the experimental study, cooler types were classified with an accuracy of over 90 %. The results showed that using machine learning algorithms, the most suitable cooler type can be selected with an accuracy of 95 % by the Extreme Gradient Boosting (XGBOOST) algorithm.

Projecting complex interactions between forest harvest and succession in the northern Acadian Forest Region

Preventing declines of native and historically-abundant tree species is an important aspect of sustainable forest management, but predicting future forest composition is challenging when succession does not tend to follow a well-defined path. We evaluated the implications of site-level interactions between timber harvesting and forest succession on the regional landscape dynamics of the complex and species-rich northern Acadian Forest Region. Our expectation was that forest composition would trend away from long-lived and shade-tolerant species, because rates of landscape disturbance from timber harvesting are high relative to historic rates of natural disturbance. We used a novel modeling approach that combined Landsat-derived time series of forest disturbance to inform realistic simulations of timber harvesting across many individual commercial forest landowners using LANDIS-II, and evaluated changes in tree species’ distributions and abundance with and without harvesting. Detailed descriptions of initial forest conditions were derived from maps of relative tree species abundance, developed using Landsat satellite imagery, regional inventory data, and an innovative machine learning algorithm. If recent harvest rates persist, simulations suggest timber harvesting will generally be sustainable in our study area; however, projected rates of site-level species turnover were high, predominantly favoring species that were less abundant under the region's natural disturbance regime. As a result, broad-scale patterns of projected species co-occurrence shifted, destabilizing important regional forest types. Our results highlight both the region's capacity for forest growth and the importance of accurately capturing the local effects of land management when projecting forested regions dominated by commercial ownership.

Integrated Geological and Geophysical Mapping of a Carbonatite-Hosting Outcrop in Siilinjärvi, Finland, Using Unmanned Aerial Systems

Mapping geological outcrops is a crucial part of mineral exploration, mine planning and ore extraction. With the advent of unmanned aerial systems (UASs) for rapid spatial and spectral mapping, opportunities arise in fields where traditional ground-based approaches are established and trusted, but fail to cover sufficient area or compromise personal safety. Multi-sensor UAS are a technology that change geoscientific research, but they are still not routinely used for geological mapping in exploration and mining due to lack of trust in their added value and missing expertise and guidance in the selection and combination of drones and sensors. To address these limitations and highlight the potential of using UAS in exploration settings, we present an UAS multi-sensor mapping approach based on the integration of drone-borne photography, multi- and hyperspectral imaging and magnetics. Data are processed with conventional methods as well as innovative machine learning algorithms and validated by geological field mapping, yielding a comprehensive and geologically interpretable product. As a case study, we chose the northern extension of the Siilinjärvi apatite mine in Finland, in a brownfield exploration setting with plenty of ground truth data available and a survey area that is partly covered by vegetation. We conducted rapid UAS surveys from which we created a multi-layered data set to investigate properties of the ore-bearing carbonatite-glimmerite body. Our resulting geologic map discriminates between the principal lithologic units and distinguishes ore-bearing from waste rocks. Structural orientations and lithological units are deduced based on high-resolution, hyperspectral image-enhanced point clouds. UAS-based magnetic data allow an insight into their subsurface geometry through modeling based on magnetic interpretation. We validate our results via ground survey including rock specimen sampling, geochemical and mineralogical analysis and spectroscopic point measurements. We are convinced that the presented non-invasive, data-driven mapping approach can complement traditional workflows in mineral exploration as a flexible tool. Mapping products based on UAS data increase efficiency and maximize safety of the resource extraction process, and reduce expenses and incidental wastes.

PRM222 - STUDY OF THE SEVERITY OF VASO-OCCLUSIVE CRISIS FOR PATIENT SUFFERING FROM SICKLE CELL DISEASE IN FRANCE USING A MACHINE LEARNING APPROACH ON THE FRENCH MEDICAL INFORMATION SYSTEM DATABASE

The PMSI (medical information system database) is the French national administrative database containing medical information and patients’ features for millions of hospital stays. Most medical studies only use descriptive methods to analyze patient cohorts. The objective of this study is to assess the capability of an innovative Machine Learning algorithm to discover hidden values in health data and to bring a better understanding of diseases. We applied this approach to predict the severity of Vaso-Occlusive Crisis (VOC) for patients suffering from Sickle-Cell Disease (SCD). The severity is expressed by the frequency of VOC and the average time between two VOC. We only selected hospital stays with an SCD related ICD-10 code. Patients hospitalized with a VOC code in 2014 were extracted and followed up for two years. 20 groups of comorbidities, 7 features related to the patient’s medical history and 8 features about the patient’s life have been identified. Data were analyzed with an Enhanced Decision Tree technique, to explain the severity of VOC with non-linear combinations of these 35 variables. 5 630 patients were included in the study. The enhanced decision tree was able to gather patients into distinct groups based on patient and disease features. 20 distinct groups were identified. The most discriminating variables for the severity of VOC were the features related to patients’ history and the number of received transfusions during the follow-up, whereas the comorbidities, the age or the gender were not discriminating. This exploratory study shows that Enhanced Decision Trees are relevant to identify patient profiles from healthcare databases and have predictive capabilities. It helps identifying leverages to prevent the severity of VOC, and so to adapt the treatment given to patients according to the risk of having a VOC in the next months.