MIMuhammad Irfan

Muhammad Irfan

*This is an algorithmically generated author record based on publicly available information.

Current Organizations / Institutes

The organizations/institutes from the scholar's latest article

Previous Organizations / Institutes

The organizations / institutes from the scholar's other articles in the past ten years


Muhammad Irfan has a total of 27 co-authors. Here are the top co-authors by number of shared publications:


Latest Publications

  • Open Access

    Smart IoT-Based Solutions for Neonatal Sleep Stratification: Single-Dual Channel EEG, AdaptiSelect, Multiview Fusion, and Rotational Ensemble Stacking

    by Muhammad Irfan, Laishuan Wang, Yan Xu, Abdulhamit Subasi, Chen Chen, Riku Klén, Tomi Westerlund, Wei Chen

    A timely diagnosis and treatment of sleep disorders in neonates during their first week of life is crucial. Current methods for staging neonatal sleep rely heavily on multiple electroencephalography (EEG) channels. These channels increase computational complexity, require a large amount of data to be transferred to the cloud, and may cause skin irritation. We propose an innovative automated classification approach that integrates multiview feature fusion, AdaptiSelect-based feature optimization, the smart cloud data transfer and reconstruction (STREAM) module, and a rotational ensemble stacking model. The data reduction module significantly enhances edge-cloud systems’ performance in IoT-based healthcare environments by reducing data transmission by a factor of 153.6 through efficient feature selection and compact data packet formation. This module ensures minimal bandwidth usage, reduces the computational load on resource-constrained edge devices, and lowers cloud storage requirements while maintaining full data reconstruction. The dataset used in this research combines two large datasets collected over four years from the Children’s Hospital Fudan University, Shanghai. A unique set of 315 features are extracted from each epoch of a single channel using flexible analytical wavelet transform (FAWT), dual-tree complex wavelet transform (DTCWT), enhanced covariance (ECOV), and spectral features based on $\alpha $ , $\beta $ , $\theta $ , and $\delta $ brain waves. These features are refined using AdaptiSelect, achieving an accuracy of 81.16% and a Kappa of 72.17% with one channel. Accuracy improves to 82.79% with a Kappa of 74.70% when using two channels, validated through ten-fold cross-validation. Additionally, Leave-One-Subject-Out cross-validation (LOSO-CV) further demonstrates the effectiveness of the proposed approach as a generalized solution. Using both single and multichannel setups, the proposed approach outperforms the most significant state-of-the-art methods in neonatal sleep analysis.


  • Multidomain Selective Feature Fusion and Stacking Based Ensemble Framework for EEG-Based Neonatal Sleep Stratification

    by Muhammad Irfan, Laishuan Wang, Husnain Shahid, Yan Xu, Abdulhamit Subasi, Adnan Munawar, Noman Mustafa, Chen Chen, Tomi Westerlund, Wei Chen

    Employing a minimal array of electroencephalography (EEG) channels for neonatal sleep stage classification is essential for data acquisition in the Internet of Medical Things (IoMT), as single-channel and edge-based features can reduce data transfer and processing requirements, enhancing cost-effectiveness and practicality. In this paper, we evaluate the efficacy of a single channel and the viability of a binary classification scheme for discerning awake and sleep states and transitions to quiet sleep. For this, two datasets of EEG signals for neonate sleep analysis were recorded from Children's Hospital of Fudan University, Shanghai, comprising recordings from 64 and 19 neonates, respectively. From each epoch, a diverse ensemble of 490 features was extracted through a blend of discrete and continuous wavelet transforms (DWT, CWT), spectral statistics, and temporal features. In addition, we introduced an innovative hybrid univariate and ensemble feature selection approach with multidomain feature fusion, a stacking-based ensemble classifier that outperforms existing work. We achieved 90.37%, 91.13%, and 94.88% accuracy for sleep/awake, quiet sleep/non-quiet sleep, and quiet sleep/awake, respectively. This was corroborated by significant Kappa values of 77.5%, 80.29%, and 89.76%. Using SelectPercentile, we devised three distinct feature selection mechanisms: one using DWT, one with CWT, and another incorporating both spectral and temporal features. Subsequently, SelectKBest was used to determine the most effective features. For our stacked model, we incorporated a trifecta of the ExtraTree model with variable estimators, a Random Forest, and an Artificial Neural Network (ANN) as base classifiers, and for the final prediction phase, ANN was implemented again. The model's performance was evaluated using K-fold and leave-one-subject cross-validation.


  • Open Access

    Single-Channel EEG Data Analysis Using a Multi-Branch CNN for Neonatal Sleep Staging

    by Hafza Ayesha Siddiqa, Zhenning Tang, Yan Xu, Laishuan Wang, Muhammad Irfan, Saadullah Farooq Abbasi, Anum Nawaz, Chen Chen, Wei Chen

    Neonatal sleep staging is crucial for understanding infant brain development and assessing neurological health. This study explores the optimal electrode configuration to reduce technical complexities and potential risks of causing skin irritation to neonates during data collection. A Multi-Branch Convolutional Neural Network (CNN) is used to categorize neonatal sleep states based on single-channel Electroencephalography (EEG) data. The proposed model was trained and tested on 16803 30-second segments from 64 infants, all of whom were at post-menstrual age between 36 and 43 weeks at the Children’s hospital of Fudan University. A total of 74 extracted time and frequency domain linear and non-linear features are applied to improve the performance of a Multi-Branch CNN-based classification model. Additionally, using principal component analysis (PCA), feature selection and feature importance are also applied to identify the most important features. Notably, the F3 channel outperforms other single-channels and has accuracy and kappa values 74.27±0.80% and 0.61, respectively. Furthermore, a combination of four left-side electrodes yields slightly better classification accuracy (75.36±0.57%) compared to the four right-side electrodes (74.76±1.10%), with corresponding kappa values of 0.63 and 0.62, respectively. In addition to providing insights into optimal electrode configuration using single-channel and multi-channel EEG data, the results highlight the critical role played by specific EEG channels in sleep stage classification. This research has the potential to enhance neonate care and monitor sleep more effectively, enabling early detection of sleep-related abnormalities such as sleep disorders. Furthermore, this research effectively captures information from a single-channel, reducing computing load while maintaining commendable performance. Additionally, integrating time and frequency domain linear and non-linear features into neonatal sleep staging can enhance accuracy and provide a deeper insight into the complex dynamics and irregularities of newborn’s sleep patterns.


  • Open Access

    A Novel Convolutional Neural Network Deep Learning Implementation for Cuffless Heart Rate and Blood Pressure Estimation

    by Géraud Bossavi, Rongguo Yan, Muhammad Irfan

    Cardiovascular diseases (CVDs) affect components of the circulatory system responsible for transporting blood through blood vessels. The measurement of the mechanical force acting on the walls of blood vessels, as well as the blood flow between heartbeats and when the heart is at rest, is known as blood pressure (BP). Regular assessment of BP can aid in the prevention and early detection of CVDs. In the present research, a deep learning algorithm was developed to accurately calculate both blood pressure (BP) and heart rate (HR) by extracting relevant features from photoplethysmogram (PPG), electrocardiogram (ECG), and ABP signals. This algorithm was implemented using the Medical Information Mart for Intensive Care (MIMIC-II) dataset. It captures vital blood pressure-related features extracted from the PPG signal and accounts for the time relationship with the ECG. The algorithm also determines the values of systolic blood pressure (SBP) and diastolic blood pressure (DBP) based on the ABP waveform through a convolutional neural network and stepwise multivariate linear regression. In comparison with other established BP measurement methods, our proposed approach achieved better results, with a mean absolute error (MAE) of approximately 4.7 mmHg for SBP and 2.1 mmHg for DBP, respectively. The standard deviation (STD) for SBP and DBP was approximately 7.6 mmHg and 3.9 mmHg, respectively. This study makes a valuable contribution to the healthcare field by introducing a novel, cost-effective continuous BP measurement method with improved accuracy while also minimizing the data dimension without losing any important information.


  • An IoT-Based Noncontact ECG System: Sole of the Feet/Hands Palm

    by Muhammad Irfan, Peng Shun, Barkoum Betra Felix, Noman Mustafa, Saadullah Farooq Abbasi, Abdelwahed Nahli, Abdulhamit Subasi, Tomi Westerlund, Wei Chen

    In smart healthcare facilities designed especially for the elderly, non-contact electrocardiogram (ECG) measurements could provide essential information about an elderly person’s health by enabling long-term health analytics. In this research work, we propose an Internet of Things (IoT) based non-contact ECG measurement system. The non-contact measurement is done using flexible electrodes that are made of fabric. These fabric-based flexible electrodes are designed to measure ECG signals from the sole of the feet (SOF) or the palms of the hands (POH) without touching human skin. To mitigate the impact of nearby electromagnetic radiation on the electrodes, a double layer of isopotential shielding is placed underneath the two active electrodes. The gathered biosignals are stored in the IoT device and transmitted to the cloud. To reduce the amount of stored and transmitted data, we improved our adaptive coding algorithm. The adaptive coding results in an average data reduction of 72%. The data can be fully recovered in the cloud for further analyses using advanced cloud-based tools in ThingSpeak. The study tested the proposed system on 35 participants, including elderly persons, adults, and children. Based on the experiments, the proposed system accurately measures the ECG signal. We validated the results with the ground truth data (polysomnography (PSG)) showing an average heart rate (HR) error of 1 beat per minute (BPM). Moreover, we compared QRS complexes detected on wrists with those detected from SOF (with or without socks), POH (with or without gloves), and one hand and one foot (with or without a sock and glove), and found no significant differences.


  • Non-Wearable IoT-Based Smart Ambient Behavior Observation System

    by Muhammad Irfan, Husnain Jawad, Barkoum Betra Felix, Saadullah Farooq Abbasi, Anum Nawaz, Saeed Akbarzadeh, Muhammad Awais, Lin Chen, Tomi Westerlund, Wei Chen

    A growing number of advanced smart systems and solutions are being designed for the elderly, helping them to live longer at home. These systems need to provide unobtrusive monitoring and safety for their users and information for the healthcare professionals and family members. Multi-modal sensor data enables the possibility for in-depth behavioral analysis. To gather multi-modal data, we propose an IoT-based smart ambient behavior observation system (SABOS). SABOS provides unobtrusive monitoring of daily living activities by utilizing various sensors integrated into the residential house. To reduce the amount of data, we present a data reduction algorithm. The data reduction algorithm effectively reduces over 90% of the submitted data with full recovery in the cloud. Data is sent to ThingSpeak for MATLAB visualization and analysis to generate graphical illustrations of daily living activities. In an emergency, an “if this then that” (IFTTT) service combined with ThingSpeak triggers an applet to send a defined message to a healthcare professional or a family member.


  • Open Access

    Crowd analysis using visual and non-visual sensors, a survey

    by Muhammad Irfan, Lucio Marcenaro, Laurissa Tokarchuk

    This paper proposes a critical survey of crowd analysis techniques using visual and non-visual sensors. Automatic crowd understanding has a massive impact on several applications including surveillance and security, situation awareness, crowd management, public space design, intelligent and virtual environments. In case of emergency, it enables practical safety applications by identifying crowd situational context information. This survey identifies different approaches as well as relevant work on crowd analysis by means of visual and non-visual techniques. Multidisciplinary research groups are addressing crowd phenomenon and its dynamics ranging from social, and psychological aspects to computational perspectives. The possibility to use smartphones as sensing devices and fuse this information with video sensors data, allows to better describe crowd dynamics and behaviors. Eventually, challenges and further research opportunities with reference to crowd analysis are exposed.