No abstract is provided for this article.
In diabetic patients presenting with macular edema (ME) shortly after cataract surgery, identifying the underlying pathology can be challenging and influence management. Our aim was to develop a simple clinical classifier able to confirm a diabetic etiology using few spectral domain optical coherence tomography parameters.We analyzed spectral domain optical coherence tomography data of 153 patients with either pseudophakic cystoid ME (n = 57), diabetic ME (n = 86), or "mixed" (n = 10). We used advanced machine learning algorithms to develop a predictive classifier using the smallest number of parameters.Most differentiating were the existence of hard exudates, hyperreflective foci, subretinal fluid, ME pattern, and the location of cysts within retinal layers. Using only 3 to 6 spectral domain optical coherence tomography parameters, we achieved a sensitivity of 94% to 98%, specificity of 94% to 95%, and an area under the curve of 0.937 to 0.987 (depending on the method) for confirming a diabetic etiology. A simple decision flowchart achieved a sensitivity of 96%, a specificity of 95%, and an area under the curve of 0.937.Confirming a diabetic etiology for edema in cases with uncertainty between diabetic cystoid ME and pseudophakic ME was possible using few spectral domain optical coherence tomography parameters with high accuracy. We propose a clinical decision flowchart for cases with uncertainty, which may support the decision for intravitreal injections rather than topical treatment.
Privacy preserving data mining aims to prevent the violation of privacy that might result from mining of sensitive data. This is commonly achieved by data anonymization. One way to anonymize data is adherence to the k-anonymity concept which requires that the probability to identify an individual by linking databases not to exceed 1/k. In this paper we propose an algorithm which utilizes rough set theory to achieve k-anonymity. The basic idea is to partition the original dataset into several disjoint reducts such that each one of them adheres to k-anonymity. We show that it is easier to make each reduct comply with k-anonymity if it does not contain all quasi-identifier attributes. Moreover, our procedure ensures that even if the attacker attempts to rejoin the reducts, the k-anonymity is still preserved.
Patients with chronic lymphocytic leukemia (CLL) have a suboptimal humoral response to vaccination. Recently, BNT162b2, an mRNA COVID-19 vaccine with a high efficacy of 95% in immunocompetent individuals, was introduced. We investigated the safety and efficacy of the BNT162b2 mRNA COVID-19 vaccine in patients with CLL from nine medical centers in Israel, Overall 400 patients were included, of whom 373 were found to be eligible for the analysis of antibody response. The vaccine appeared to be safe and only grade 1-2 adverse events were seen in 50% of the patients. Following the second dose, an antibody response was detected in 43% of the cohort. Among these CLL patients, 61% of the treatment-na ve patients responded to the vaccine, while responses developed in only 18% of those with ongoing disease, 37% of those previously treated with a BTK inhibitor and 5% of those recently given an anti-CD20 antibody. Among patients treated with BCL2 as monotherapy or in combination with anti-CD20, 62% and 14%, respectively, developed an immune response. There was a high concordance between neutralizing antibodies and positive serological response to spike protein. Based on our findings we developed a simple seven-factor score including timing of any treatment with anti-CD20, age, treatment status, and IgG, IgA, IgM and hemoglobin levels. The sum of all the above parameters can serve as a possible estimate to predict whether a given CLL patient will develop sufficient antibodies. In conclusion, the BNT162b2 mRNA COVID-19 vaccine was found to be safe in patients with CLL, but its efficacy is limited, particularly in treated patients.
Gradient boosting models like XGBoost are among the most popular models for tabular classification problems. Unfortunately, the greediness of gradient boosting algorithms can cause them to rely too heavily on some features, thereby starving the other features. We propose Iterative Feature eXclusion (IFX) to alleviate this problem by iteratively removing the most influential feature from the training data and continuing training. By forcing the model to learn from weaker features, we increase the diversity of the gradient boosting model and improve the predictive performance. Our experiments show that in most cases, IFX improves XGBoost predictive performance, sometimes by a large margin. All of the code and results from our experiments are freely available online. Iterative Feature eXclusion can be used as a drag-and-drop replacement for XGBoost, thereby easing the adoption of our work by machine learning researchers and practitioners.
We present a new concept - Wikiometrics - the derivation of metrics and indicators from Wikipedia. Wikipedia provides an accurate representation of the real world due to its size, structure, editing policy and popularity. We demonstrate an innovative mining methodology, where different elements of Wikipedia - content, structure, editorial actions and reader reviews - are used to rank items in a manner which is by no means inferior to rankings produced by experts or other methods. We test our proposed method by applying it to two real-world ranking problems: top world universities and academic journals. Our proposed ranking methods were compared to leading and widely accepted benchmarks, and were found to be extremely correlative but with the advantage of the data being publically available.
In previous work we proposed the M-score measure for assigning a misuseability (i.e., sensitivity) score to data records. The M-score uses sensitivity score functions that should be acquired from domain experts. In this paper we present two different approaches for acquiring the required knowledge. In the first method the expert is asked to explicitly assign a sensitivity score to displayed records. The second method employs pairwise comparison approach. A field study indicates that the later method is preferable.
No abstract is provided for this article.
Despite continuous investments in data technologies, the latency of querying data still poses a significant challenge. Modern analytic solutions require near real-time responsiveness both to make them interactive and to support automated processing. Current technologies (Hadoop, Spark, Dataflow) scan the dataset to execute queries and focus on providing scalable data storage and in-memory concurrent data processing to maximize task execution speed. We argue that these solutions fail to offer an adequate level of interactivity, since they depend on continual access to data. In this paper, we present a method for query approximation, also known as approximate query processing (AQP), that reduces the need to scan data during inference (query calculation), thus enabling a rapid query processing tool. We use an LSTM network to learn the relationship between queries and their results, and to provide a rapid inference layer for the prediction of query results. Our method (referred to as "Hunch") produces a lightweight LSTM network which provides high query throughput. We evaluated our method using 12 datasets and compared it to state-of-the-art AQP engines (VerdictDB, BlinkDB) in terms of the query latency, model weight, and accuracy. The results show that our method predicted query results with a normalized root mean squared error (NRMSE) ranging from approximately 1% to 4%, which, for the majority of our datasets, was better than the results of the benchmarks. Moreover, our method was able to predict up to 120,000 queries in a second (streamed together) and with a single query latency of no more than 2 ms.
No abstract is provided for this article.
Objective To use machine learning-based methods in designing a predictive model of rehabilitation outcomes for postacute hip fracture patients. Design A retrospective analysis using linear models, AdaBoost, CatBoost, ExtraTrees, K-Nearest Neighbors, RandomForest, Support vector machine, XGBoost, and voting of all models to develop and validate a predictive model. Setting A university-affiliated 300-bed major postacute geriatric rehabilitation center. Participants Consecutive hip fracture patients (N=1625) admitted to an postacute rehabilitation department. Main Outcome Measures The FIM instrument, motor FIM (mFIM), and the relative functional gain on mFIM (mFIM effectiveness) as a continuous and binary variable. Ten predictive models were created: base models (linear/logistic regression), and 8 machine learning models (AdaBoost, CatBoost, ExtraTrees, K-Nearest Neighbors, RandomForest, Support vector machine, XGBoost, and a voting ensemble). R 2 was used to evaluate their performance in predicting a continuous outcome variable, and the area under the receiver operating characteristic curve was used to evaluate the binary outcome. A paired 2-tailed t test compared the results of the different models. Results Machine learning-based models yielded better results than the linear and logistic regression models in predicting rehabilitation outcomes. The 3 most important predictors of the mFIM effectiveness score were the Mini Mental State Examination (MMSE), prefracture mFIM scores, and age. The 3 most important predictors of the discharge mFIM score were the admission mFIM, MMSE, and prefracture mFIM scores. The most contributing factors for favorable outcomes (mFIM effectiveness > median) with higher prediction confidence level were high MMSE (25.7±2.8), high prefacture mFIM (81.5±7.8), and high admission mFIM (48.6±8) scores. We present a simple prediction instrument for estimating the expected performance of postacute hip fracture patients. Conclusions The use of machine learning models to predict rehabilitation outcomes of postacute hip fracture patients is superior to linear and logistic regression models. The higher the MMSE, prefracture mFIM, and admission mFIM scores are, the higher the confidence levels of the predicted parameters.
Driving under the influence of alcohol is a widespread phenomenon in the US where it is considered a major cause of fatal accidents. In this research we present a novel approach and concept for detecting intoxication from motion differences obtained by the sensors of wearable devices. We formalize the problem of drunkenness detection as a supervised machine learning task, both as a binary classification problem (drunk or sober) and a regression problem (the breath alcohol content level). In order to test our approach, we collected data from 30 different subjects (patrons at three bars) using Google Glass and the LG G-watch, Microsoft Band, and Samsung Galaxy S4. We validated our results against an admissible breathalyzer used by the police. A system based on this concept, successfully detected intoxication and achieved the following results: 0.95 AUC and 0.05 FPR, given a fixed TPR of 1.0. Applications based on our system can be used to analyze the free gait of drinkers when they walk from the car to the bar and vice-versa, in order to alert people, or even a connected car and prevent people from driving under the influence of alcohol.
