A machine learning based framework for IoT devices identification using web traffic

Identification of the Internet of Things (IoT) devices has become an essential part of network management to secure the privacy of smart homes and offices. With its wide adoption in the current era, IoT has facilitated the modern age in many ways. However, such proliferation also has associated privacy and data security risks. In the case of smart homes and smart offices, unknown IoT devices increase vulnerabilities and chances of data theft. It is essential to identify the connected devices for secure communication. It is very difficult to maintain the list of rules when the number of connected devices increases and human involvement is necessary to check whether any intruder device has approached the network. Therefore, it is required to automate device identification using machine learning methods. In this article, we propose an accuracy boosting model (ABM) using machine learning models of random forest and extreme gradient boosting. Featuring engineering techniques are employed along with cross-validation to accurately identify IoT devices such as lights, smoke detectors, thermostat, motion sensors, baby monitors, socket, TV, security cameras, and watches. The proposed ensemble model utilizes random forest (RF) and extreme gradient boosting (XGB) as base learners with adaptive boosting. The proposed ensemble model is tested with extensive experiments involving the IoT Device Identification dataset from a public repository. Experimental results indicate a higher accuracy of 91%, precision of 93%, recall of 93%, and F1 score of 93%.

Predicting Thalassemia Using Feature Selection Techniques: A Comparative Analysis

Thalassemia represents one of the most common genetic disorders worldwide, characterized by defects in hemoglobin synthesis. The affected individuals suffer from malfunctioning of one or more of the four globin genes, leading to chronic hemolytic anemia, an imbalance in the hemoglobin chain ratio, iron overload, and ineffective erythropoiesis. Despite the challenges posed by this condition, recent years have witnessed significant advancements in diagnosis, therapy, and transfusion support, significantly improving the prognosis for thalassemia patients. This research empirically evaluates the efficacy of models constructed using classification methods and explores the effectiveness of relevant features that are derived using various machine-learning techniques. Five feature selection approaches, namely Chi-Square (χ2), Exploratory Factor Score (EFS), tree-based Recursive Feature Elimination (RFE), gradient-based RFE, and Linear Regression Coefficient, were employed to determine the optimal feature set. Nine classifiers, namely K-Nearest Neighbors (KNN), Decision Trees (DT), Gradient Boosting Classifier (GBC), Linear Regression (LR), AdaBoost, Extreme Gradient Boosting (XGB), Random Forest (RF), Light Gradient Boosting Machine (LGBM), and Support Vector Machine (SVM), were utilized to evaluate the performance. The χ2 method achieved accuracy, registering 91.56% precision, 91.04% recall, and 92.65% f-score when aligned with the LR classifier. Moreover, the results underscore that amalgamating over-sampling with Synthetic Minority Over-sampling Technique (SMOTE), RFE, and 10-fold cross-validation markedly elevates the detection accuracy for αT patients. Notably, the Gradient Boosting Classifier (GBC) achieves 93.46% accuracy, 93.89% recall, and 92.72% F1 score.

Clinicopathological spectrum of Diffuse Large B Cell lymphoma: a study targeting population yet unexplored in Pakistan

Objective

Diffuse Large B Cell Lymphoma (DLBCL) is the most common type of Non-Hodgkin Lymphoma (NHL). The aim of this study was to assess the clinico pathological characteristics of DLBCL specifically, among the affected individuals residing in Northern areas of Pakistan who had not been previously included in major lymphoma studies due to their remote location.

Results

Mean age of the patients was 49.7 years. Male: female ratio was 1.5:1. Primary site was lymph node in 99 (71.74%) patients, out of which, 36 (26.09%) patients had B symptoms and 19 (13.77%) patients had stage IV disease. 39 (28.26%) patients had primary extra nodal involvement, 4 (2.90%) patients had B symptoms and 3 (2.17%) had stage IV disease. Extra nodal sites involved in primary extra nodal DLBCL were gastrointestinal tract (GIT) 19 (48.72%), tonsils 6 (15.38%), spine 4 (10.26%), soft tissue swelling 3 (7.69%), parotid gland 2 (5.13%), thyroid 2 (5.13%) central nervous system (CNS) 1 (2.56), breast 1 (2.56%) and bone marrow 1 (2.56%). Our study revealed increased percentage of patients with nodal DLBCL in stage IV and with B symptoms. Few patients with primary extra nodal DLBCL had B symptoms and stage IV disease at presentation. GIT was the most common site of involvement in primary extra nodal DLBCL.

Supplementary Information

The online version contains supplementary material available at 10.1186/s13104-021-05768-5.

A performance comparison of supervised machine learning models for Covid-19 tweets sentiment analysis

The spread of Covid-19 has resulted in worldwide health concerns. Social media is increasingly used to share news and opinions about it. A realistic assessment of the situation is necessary to utilize resources optimally and appropriately. In this research, we perform Covid-19 tweets sentiment analysis using a supervised machine learning approach. Identification of Covid-19 sentiments from tweets would allow informed decisions for better handling the current pandemic situation. The used dataset is extracted from Twitter using IDs as provided by the IEEE data port. Tweets are extracted by an in-house built crawler that uses the Tweepy library. The dataset is cleaned using the preprocessing techniques and sentiments are extracted using the TextBlob library. The contribution of this work is the performance evaluation of various machine learning classifiers using our proposed feature set. This set is formed by concatenating the bag-of-words and the term frequency-inverse document frequency. Tweets are classified as positive, neutral, or negative. Performance of classifiers is evaluated on the accuracy, precision, recall, and F1 score. For completeness, further investigation is made on the dataset using the Long Short-Term Memory (LSTM) architecture of the deep learning model. The results show that Extra Trees Classifiers outperform all other models by achieving a 0.93 accuracy score using our proposed concatenated features set. The LSTM achieves low accuracy as compared to machine learning classifiers. To demonstrate the effectiveness of our proposed feature set, the results are compared with the Vader sentiment analysis technique based on the GloVe feature extraction approach.

UCDiff: Difference Detection in Use Case Models of Healthcare System

Development of large-scale healthcare software projects essentially need the efficient management of the created software artifacts during software development process. In such projects different versions of an artifact are created at different times. Traditional software configuration management systems, such as Git, Subversion (SVN), etc., are designed for later phases of software development, which mainly handle the source code document. These systems are unable to perform difference detection and version management tasks on models such as unified modeling language diagrams. UML use case model is used for capturing functional requirements at analyses phase. Different versions of the use case model are created during analyses phase. This paper addresses the detection of differences between two versions of a use case model. In order to perform difference detection, we need to perform three main tasks, i. e., extract the contents of the model, comparison of models and difference representation. Most of the existing approaches in literature of model comparison deal with UML class diagrams. To the best of our knowledge, so far no appropriate approach addresses difference computation of use case model. Existing approaches are not applicable on use case model due to different semantics of use case model. In this research, the concept of model-based software configuration management (SCM) for use case difference detection is proposed. The use case models are created in an open source tool, starUML. The proposed difference algorithm is applied on intermediate tree structure representation of models. As a case study, different versions of a patient appointment healthcare system is used to evaluate different evaluation parameters, such as accuracy, domain independence, high conceptual level and tool independence.

A Feature-Based Evaluation of Model Merge Methods for e-Health Solutions

Model-driven engineering (MDE) paradigm considers models as central artifacts for software development lifecycle during which models evolve. Developing an e-health solution using MDE poses challenges of model version control, model differencing and model merging, which requires appropriate software configuration management (SCM). In this paper we focus on model-driven merging, which refers to combining two or more versions of a model into a single consolidated version. SCM for model-driven merging leverages evolution of valid configurations, which is a highly desired behavior. Our investigation is based on the features that are required for model-driven SCM realization. Initially, we identify these features using which the existing model-driven merging techniques are evaluated. It is observed that though various proposals are made by academia and research community, a standard model-driven SCM solution that can cater to the needs of industry is still absent. This is in contrary to the situation of traditional SCM systems where standard solutions exist. We also present the usefulness of each technique along with the tradeoffs involved. Finally, guidelines are provided to select techniques appropriate for given circumstances.

A Feature-Based Evaluation of Model Merge Methods for e-Health Solutions

Model-driven engineering (MDE) paradigm considers models as central artifacts for software development lifecycle during which models evolve. Developing an e-health solution using MDE poses challenges of model version control, model differencing and model merging, which requires appropriate software configuration management (SCM). In this paper we focus on model-driven merging, which refers to combining two or more versions of a model into a single consolidated version. SCM for model-driven merging leverages evolution of valid configurations, which is a highly desired behavior. Our investigation is based on the features that are required for model-driven SCM realization. Initially, we identify these features using which the existing model-driven merging techniques are evaluated. It is observed that though various proposals are made by academia and research community, a standard model-driven SCM solution that can cater to the needs of industry is still absent. This is in contrary to the situation of traditional SCM systems where standard solutions exist. We also present the usefulness of each technique along with the tradeoffs involved. Finally, guidelines are provided to select techniques appropriate for given circumstances.

UCDiff: Difference Detection in Use Case Models of Healthcare System

Development of large-scale healthcare software projects essentially need the efficient management of the created software artifacts during software development process. In such projects different versions of an artifact are created at different times. Traditional software configuration management systems, such as Git, Subversion (SVN), etc., are designed for later phases of software development, which mainly handle the source code document. These systems are unable to perform difference detection and version management tasks on models such as unified modeling language diagrams. UML use case model is used for capturing functional requirements at analyses phase. Different versions of the use case model are created during analyses phase. This paper addresses the detection of differences between two versions of a use case model. In order to perform difference detection, we need to perform three main tasks, i. e., extract the contents of the model, comparison of models and difference representation. Most of the existing approaches in literature of model comparison deal with UML class diagrams. To the best of our knowledge, so far no appropriate approach addresses difference computation of use case model. Existing approaches are not applicable on use case model due to different semantics of use case model. In this research, the concept of model-based software configuration management (SCM) for use case difference detection is proposed. The use case models are created in an open source tool, starUML. The proposed difference algorithm is applied on intermediate tree structure representation of models. As a case study, different versions of a patient appointment healthcare system is used to evaluate different evaluation parameters, such as accuracy, domain independence, high conceptual level and tool independence.

Home automation using general purpose household electric appliances with Raspberry Pi and commercial smartphone

This study presents the design and implementation of a home automation system that focuses on the use of ordinary electrical appliances for remote control using Raspberry Pi and relay circuits and does not use expensive IP-based devices. Common Lights, Heating, Ventilation, and Air Conditioning (HVAC), fans, and other electronic devices are among the appliances that can be used in this system. A smartphone app is designed that helps the user to design the smart home to his actual home via easy and interactive drag & drop option. The system provides control over the appliances via both the local network and remote access. Data logging over the Microsoft Azure cloud database ensures system recovery in case of gateway failure and data record for lateral use. Periodical notifications also help the user to optimize the usage of home appliances. Moreover, the user can set his preferences and the appliances are auto turned off and on to meet user-specific requirements. Raspberry Pi acting as the server maintains the database of each appliance. HTTP web interface and apache server are used for communication between the android app and raspberry pi. With a 5v relay circuit and micro-processor Raspberry Pi, the proposed system is low-cost, energy-efficient, easy to operate, and affordable for low-income houses.

Systematic case finding for tuberculosis in HIV-infected people who inject drugs: experience from Pakistan

SETTING

Pakistan is a high tuberculosis (TB) burden country, moving from low human immunodeficiency virus (HIV) prevalence to a concentrated epidemic driven primarily by people who inject drugs (PWID). The Antiretroviral Treatment Adherence Unit (AAU) in Islamabad, Pakistan, is a residential facility that offers combined treatment for opioid dependence and HIV.

OBJECTIVE AND DESIGN

This retrospective study was conducted to assess TB prevalence among HIV-infected PWID referred to the AAU and to evaluate the diagnostic value of cough as a screening symptom. A single sputum sample was collected regardless of symptoms, and examined using smear, Xpert® MTB/RIF and culture.

RESULTS

Of 888 PWID, 71.5% submitted a sputum sample. More TB cases were detected using Xpert (n = 25) than with smear (n = 10) or culture (n = 20). A TB prevalence of 6141 per 100 000 was estimated based on seven cases already identified as being on anti-tuberculosis treatment and 32 newly diagnosed bacteriologically confirmed TB cases. Both cough and smoking (10 pack-years) were associated with increased TB prevalence. Only half of the TB cases reported cough. Rifampicin resistance was reported among 10% (3/29) of newly identified cases.

CONCLUSION

TB prevalence in HIV-infected PWID was 15 times higher than in the general adult population. As a screening symptom, cough has low diagnostic value.

The benefit of trans-attic endoscopic control of ossicular prosthesis after cholesteatoma surgery

OBJECTIVE

To show the efficiency of using transmastoid atticotomy (TMA) endoscopy on the outcome of ossiculoplasty in patients with cholesteatoma. TMA is often performed as part of the surgical management of patients with middle ear cholesteatoma extending to the epitympanum. TMA can also be used as an access for endoscopic view to confirm the right alignment and stability of the ossicular prosthesis because the reconstruction of the tympanic membrane will obscure the visualization of the prosthesis.

METHODS

A retrospective study was done at a tertiary referral institute, including 133 ears with cholesteatoma that underwent canal wall-up tympanomastoidectomy (CWU) with ossicular reconstruction using titanium prosthesis between August 2013 and August 2015. Post packing of the ear canal and position, stability, and axis of the prosthesis were checked using endoscope positioned in the attic through TMA. A postoperative pure-tone average air-bone gap (ABG) of 20 dB or less was considered as a successful hearing result. Results are compared with historical control groups.

RESULTS

Of the 133 ears, 88 patients underwent reconstruction with partial ossicular replacement prosthesis (PORP), whereas the rest (45 patients) had total ossicular replacement prosthesis (TORP). A postoperative ABG ≤ 20 dB was obtained in 77.4% of all the patients (79.5% for PORP; 73.3% for TORP).

CONCLUSION

Endoscopic assessment of the ossicular prosthesis via the attic, after repositioning of the tympanomeatal flap and packing the ear canal, decreases the risk of immediate ossiculoplasty failure and improves the functional outcome after ossicular chain reconstruction in cholesteatoma surgery.

LEVEL OF EVIDENCE

4 Laryngoscope, 129:2754-2759, 2019.

Discovery and classification of user interests on social media

Purpose

Twitter users’ generated data, known as tweets, are now not only used for communication and opinion sharing, but they are considered an important source of trendsetting, future prediction, recommendation systems and marketing. Using network features in tweet modeling and applying data mining and deep learning techniques on tweets is gaining more and more interest.

Design/methodology/approach

In this paper, user interests are discovered from Twitter Trends using a modeling approach that uses network-based text data (tweets). First, the popular trends are collected and stored in separate documents. These data are then pre-processed, followed by their labeling in respective categories. Data are then modeled and user interest for each Trending topic is calculated by considering positive tweets in that trend, average retweet and favorite count.

Findings

The proposed approach can be used to infer users’ topics of interest on Twitter and to categorize them. Support vector machine can be used for training and validation purposes. Positive tweets can be further analyzed to find user posting patterns. There is a positive correlation between tweets and Google data.

Practical implications

The results can be used in the development of information filtering and prediction systems, especially in personalized recommendation systems.

Social implications

Twitter microblogging platform offers content posting and sharing to billions of internet users worldwide. Therefore, this work has significant socioeconomic impacts.

Originality/value

This study guides on how Twitter network structure features can be exploited in discovering user interests using tweets. Further, positive correlation of Twitter Trends with Google Trends is reported, which validates the correctness of the authors’ approach.

Gastric specimens for diagnosing tuberculosis in adults unable to expectorate in Rawalpindi, Pakistan

Setting: Adult pulmonary tuberculosis (TB) patients unable to expectorate quality sputum represent a diagnostic challenge. A private hospital in Pakistan routinely performs gastric aspiration in adults with difficulties expectorating. Objective: To assess the usefulness of gastric specimens (GS) in diagnosing pulmonary TB (PTB) and drug-resistant TB in adult presumptive TB patients unable to expectorate, and to compare the diagnostic yield and sensitivity of smear, culture and the Xpert^® MTB/RIF assay. Design: This was a comparative cross-sectional study based on retrospective record review. Results: Of 900, 885 and 877 GS tested by smear, Xpert and culture, respectively, interpretable results were obtained for respectively 900 (100%), 859 (97.1%) and 754 (86.0%), with a diagnostic yield of respectively 23.6%, 30.3% and 24.9%. The yield was significantly higher for Xpert in previously treated patients. There were 313 patients with definite TB, defined as positive on Xpert and/or culture. The 82.8% sensitivity of Xpert was significantly higher than that of smear (61.0%) and culture (67.8%). Conclusion: GS obtained by aspiration under routine programme conditions is useful for detecting TB and drug-resistant TB in adult patients unable to expectorate. Xpert, with its rapid testing, high proportion of interpretable results and better sensitivity, can substantially improve the diagnosis of bacteriologically confirmed TB and rifampicin resistance.

Preoperative Predictors in Chronic Suppurative Otitis Media for Ossicular Chain Discontinuity: A Cross-Sectional Study

OBJECTIVE

Ossicular discontinuity may result from chronic suppurative otitis media and is usually detected intraoperatively. Our objective is to determine whether a preoperative audiogram can preoperatively predict the presence or absence of ossicular discontinuity.

METHODS

A cross-sectional study was prospectively run on our patients, aged 12-75 years, ultimately operated on for chronic suppurative otitis media. Preoperative audiograms were analyzed to measure frequency-specific air-bone gap (ABG) cutoff values. Intraoperatively, ossicular chain integrity was carefully checked. Logistic regression analysis was done to obtain a predictive model.

RESULTS

A total of 270 patients (306 ears) were included. Frequency-specific ABG cutoff values can predict ossicular discontinuity, namely: high ABGs at 1,000 Hz (>27.5 dB) and 2,000 Hz (>17.5 dB) are the most reliable variables associated with ossicular discontinuity.

CONCLUSION

Preoperative audiograms can predict the presence of ossicular discontinuity in chronic suppurative otitis media. Large ABGs at both 1,000 and 2,000 Hz can predict ossicular discontinuity with a great degree of certainty.

Chemical Closure of Tympanic Membrane Perforation: Call for Caution

Chemical closure of tympanic membrane perforation is a commonly practiced office-based otological procedure, which is labeled to be effective and safe. In this paper, we report a case of a young lady with disastrous complications following an attempt of chemical cauterization of her perforated tympanic membrane.