Accurate estimation of dental age is crucial in forensic and clinical applications; however, traditional methods often suffer from subjectivity and observer bias. This study investigates the performance of advanced machine learning (ML) algorithms for estimating dental age based on features extracted from ConeBeam Computed Tomography (CBCT) images of mandibular canines and premolars in the Jordanian population. Eleven features encompassing volumetric measurements, morphological dimensions, and categorical dental aging stages were analyzed. Models were rigorously assessed using $\mathbf{1 0}$-fold cross-validation across male, female, and combined datasets. CatBoost and Gradient Boosting regressors consistently outperformed conventional regression methods, demonstrating superior predictive accuracy as measured by Mean Absolute Error (MAE), Root Mean Squared Error (RMSE), and Standard Error of Estimate (SEE). Notably, the CatBoost regressor exhibited exceptional robustness, achieving the best overall performance with a MAE of 7.14 ± 2.11 (Male), $6.19 \pm 1.95$ (Female), and $6.64 \pm 1.15$ (combined). Additionally, the Support Vector Regressor achieved the lowest MAE for females: $6.04 \pm 2.45$. Decision Tree and Gradient Boosting models also demonstrated commendable accuracy, further emphasizing the effectiveness of ensemble-based ML approaches. These results highlight the strong potential of ML techniques to enhance objectivity, accuracy, and applicability in dental age estimation by leveraging comprehensive dental feature sets.

One of the aims of population genetics is to identify genetic differences/similarities among individuals of multiple ancestries. Many approaches including principal component analysis, clustering, and maximum likelihood techniques can be used to assign individuals to a given ancestry based on their genetic makeup. Although there are several tools that implement such algorithms, there is a lack of interactive visual platforms to run a variety of algorithms in one place. Therefore, we developed PopMLvis, a platform that offers an interactive environment to visualize genetic similarity data using several algorithms, and generate figures that can be easily integrated into scientific articles.

This demo presents GTI; a graph-based trajectory imputation framework that aims to impute sparse trajectory datasets to boost their accuracy. GTI can act as a pre-processing step to increase the accuracy of any trajectory data management system or trajectory-based application. Unlike the large majority of existing trajectory imputation frameworks, GTI assumes that the underlying road network is not available. Audience will be able to interact with GTI through different scenarios that show how GTI can be used and customized to improve the quality of trajectory data in their corresponding spatial and temporal aspects.

GPS-enabled devices, including vehicles, smartphones, wearable and tracking devices, as well as various check-in and social network data are continuously producing tremendous amounts of trajectory data, which are used consistently in many applications such as urban planning and map inference. Existing techniques for trajectory data imputation rely heavily on the existing maps to perform map-matching operations. However, modern applications such as map construction and map update assume no map exists. In this paper, we propose GTI - a scalable graph-based trajectory imputation approach for trajectory data completion. GTI relies on cross-trajectory imputation, as it exploits "mutual information" of the aggregated knowledge of all input sparse trajectories to impute the missing data for each single one of them. GTI can act as a pre-processing step for any trajectory data management system or trajectory-based application, as it takes raw sparse trajectory data as its input and outputs dense imputed trajectory data that significantly increase the accuracy of different systems that consume trajectory data. We evaluate GTI on junction-scale as well as city-scale real datasets. In addition, GTI is used as a pre-processing step in multiple trajectory-based applications and it boosts the accuracy across these applications compared with the state-of-the-art work.

The ability to collect large numbers of trajectory data through GPS-enabled devices have enabled a myriad of very important applications that are widely used on a daily basis. This includes urban computing, transportation, and map APIs for routing and navigation. Unfortunately, a major hinder for all these applications is the accuracy of collected trajectories. Due to low sampling rates, trajectories are usually sparse in terms of the large spatial and temporal distances between each two consecutive collected points. This paper presents TrImpute; a novel framework for trajectory imputation that inserts artificial GPS points between the real ones in a way that the imputed trajectories end up to be very similar to the case if such trajectories were collected with a much higher sampling rate. Unlike all prior trajectory imputation techniques, TrImpute does not assume the knowledge of the underlying road network. This makes it more practical when the underlying road network is not available or inaccurate. Experimental results on real datasets and a real deployment of TrImpute show that it is highly scalable, accurate, and can significantly boost the performance of trajectory applications by feeding them highly accurate trajectories.

This demo presents QARTA; an open-source full-fledged system for highly accurate and scalable map services. QARTA employs machine learning techniques to: (a) construct its own highly accurate map in terms of both map topology and edge weights, and (b) calibrate its query answers based on contextual information, including transportation modality, underlying algorithm, and time of day/week. The demo is based on actual deployment of QARTA in all Taxis in the State of Qatar and in the third-largest food delivery company in the country, and receiving hundreds of thousands of daily API calls with a real-time response time. Audience will be able to interact with the demo through various scenarios that show QARTA map and query accuracy as well as internals of QARTA.

During heart failure (HF) at the cellular level, the electrophysiological properties of single myocytes get remodeled, which can trigger the occurrence of ventricular arrhythmias that could be manifested in many forms such as early afterdepolarizations (EADs) and alternans (ALTs). In this paper, based on experimentally observed human HF data, specific ionic and exchanger current strengths are modified from a recently developed human ventricular cell model: the O'Hara-Virág-Varró-Rudy (OVVR) model. A new transmural HF-OVVR model is developed that incorporates HF changes and variability of the observed remodeling. This new heterogeneous HF-OVVR model is able to replicate many of the failing action potential (AP) properties and the dynamics of both [Ca(2+)]i and [Na(+)]i in accordance with experimental data. Moreover, it is able to generate EADs for different cell types and exhibits ALTs at modest pacing rate for transmural cell types. We have assessed the HF-OVVR model through the examination of the AP duration and the major ionic currents' rate dependence in single myocytes. The evaluation of the model comes from utilizing the steady-state (S-S) and S1-S2 restitution curves and from probing the accommodation of the HF-OVVR model to an abrupt change in cycle length. In addition, we have investigated the effect of chosen currents on the AP properties, such as blocking the slow sodium current to shorten the AP duration and suppress the EADs, and have found good agreement with experimental observations. This study should help elucidate arrhythmogenic mechanisms at the cellular level and predict unseen properties under HF conditions. In addition, this AP cell model might be useful for modeling and simulating HF at the tissue and organ levels.