2024
Computers in Biology and Medicine (Peer-reviewed)
Open Access Elsevier Journal (Impact Factor = 7.7)
Article Title
"An ensemble machine learning approach for the detection of unannounced meals to enhance postprandial glucose control"
Authors
Muhammad Ibrahim, Aleix Beneyto, Ivan Contreras, and Josep Vehi
Publication date & DOI
February, 2024 | DOI: https://doi.org/10.1016/j.compbiomed.2024.108154
Abstract
Background: Hybrid automated insulin delivery systems enhance postprandial glucose control in type 1 diabetes, however, meal announcements are burdensome. To overcome this, we propose a machine learning-based automated meal detection approach;
Methods: A heterogeneous ensemble method combining an artificial neural network, random forest, and logistic regression was employed. Trained and tested on data from two in-silico cohorts comprising 20 and 47 patients. It accounted for various meal sizes (moderate to high) and glucose appearance rates (slow and rapid absorbing). To produce an optimal prediction model, three ensemble configurations were used: logical AND, majority voting, and logical OR. In addition to the in-silico data, the proposed meal detector was also trained and tested using the OhioT1DM dataset. Finally, the meal detector is combined with a bolus insulin compensation scheme;
Results: The ensemble majority voting obtained the best meal detector results for both the in-silico and OhioT1DM cohorts with a sensitivity of 77%, 94%, 61%, precision of 96%, 89%, 72%, F1-score of 85%, 91%, 66%, and with false positives per day values of 0.05, 0.19, 0.17, respectively. Automatic meal detection with insulin compensation has been performed in open-loop insulin therapy using the AND ensemble, chosen for its lower false positive rate. Time-in-range has significantly increased 10.48% and 16.03%, time above range was reduced by 5.16% and 11.85%, with a minimal time below range increase of 0.35% and 2.69% for both in-silico cohorts, respectively, compared to the results without a meal detector;
Conclusion: To increase the overall accuracy and robustness of the predictions, this ensemble methodology aims to take advantage of each base model’s strengths. All of the results point to the potential application of the proposed meal detector as a separate module for the detection of meals in automated insulin delivery systems to achieve improved glycemic control.
"An ensemble machine learning approach for the detection of unannounced meals to enhance postprandial glucose control"
Authors
Muhammad Ibrahim, Aleix Beneyto, Ivan Contreras, and Josep Vehi
Publication date & DOI
February, 2024 | DOI: https://doi.org/10.1016/j.compbiomed.2024.108154
Abstract
Background: Hybrid automated insulin delivery systems enhance postprandial glucose control in type 1 diabetes, however, meal announcements are burdensome. To overcome this, we propose a machine learning-based automated meal detection approach;
Methods: A heterogeneous ensemble method combining an artificial neural network, random forest, and logistic regression was employed. Trained and tested on data from two in-silico cohorts comprising 20 and 47 patients. It accounted for various meal sizes (moderate to high) and glucose appearance rates (slow and rapid absorbing). To produce an optimal prediction model, three ensemble configurations were used: logical AND, majority voting, and logical OR. In addition to the in-silico data, the proposed meal detector was also trained and tested using the OhioT1DM dataset. Finally, the meal detector is combined with a bolus insulin compensation scheme;
Results: The ensemble majority voting obtained the best meal detector results for both the in-silico and OhioT1DM cohorts with a sensitivity of 77%, 94%, 61%, precision of 96%, 89%, 72%, F1-score of 85%, 91%, 66%, and with false positives per day values of 0.05, 0.19, 0.17, respectively. Automatic meal detection with insulin compensation has been performed in open-loop insulin therapy using the AND ensemble, chosen for its lower false positive rate. Time-in-range has significantly increased 10.48% and 16.03%, time above range was reduced by 5.16% and 11.85%, with a minimal time below range increase of 0.35% and 2.69% for both in-silico cohorts, respectively, compared to the results without a meal detector;
Conclusion: To increase the overall accuracy and robustness of the predictions, this ensemble methodology aims to take advantage of each base model’s strengths. All of the results point to the potential application of the proposed meal detector as a separate module for the detection of meals in automated insulin delivery systems to achieve improved glycemic control.
2021
IEEE Access (Peer-reviewed)
Open Access Journal (Impact Factor = 3.9)
Article Title
"An FPGA-Based Accelerated Mutation Detection System for the Tumor Suppressor Gene"
Authors
Muhammad Ibrahim, Omer Mujahid, Najib ur Rehman, Azhar Qazi, Zahid Ullah, Tama Fouzder
Publication date & DOI
December, 2021 | DOI: 10.1109/ACCESS.2021.3134284
Abstract
This paper proposes a novel fast mutation detection system that looks for mutations in the tumor suppressor gene, also known as TP53. Mutations are modifications in the nucleotide sequences of the human genome and may be caused by various factors, such as exposure to radiation, sunlight, smoking and replication errors. Mutations in TP53 are the most common cause of cancer and early detection may prevent cancer from happening. The proposed system utilizes the high matching speed of a logic-based content-addressable memory (CAM) for mutation detection along with a hamming distance calculator that specifies the exact location of the mutation. The proposed system implementation is carried on a Xilinx Virtex®-7field-programmable gate array (FPGA) and demonstrates a low match time of 0.18 μs, which is much faster compared to the state-of-the-art systems.
"An FPGA-Based Accelerated Mutation Detection System for the Tumor Suppressor Gene"
Authors
Muhammad Ibrahim, Omer Mujahid, Najib ur Rehman, Azhar Qazi, Zahid Ullah, Tama Fouzder
Publication date & DOI
December, 2021 | DOI: 10.1109/ACCESS.2021.3134284
Abstract
This paper proposes a novel fast mutation detection system that looks for mutations in the tumor suppressor gene, also known as TP53. Mutations are modifications in the nucleotide sequences of the human genome and may be caused by various factors, such as exposure to radiation, sunlight, smoking and replication errors. Mutations in TP53 are the most common cause of cancer and early detection may prevent cancer from happening. The proposed system utilizes the high matching speed of a logic-based content-addressable memory (CAM) for mutation detection along with a hamming distance calculator that specifies the exact location of the mutation. The proposed system implementation is carried on a Xilinx Virtex®-7field-programmable gate array (FPGA) and demonstrates a low match time of 0.18 μs, which is much faster compared to the state-of-the-art systems.
2023
22nd IFAC World Congress (Peer-reviewed)
International Federation of Automatic Control (IFAC) 22nd World Congress, July 9-14, 2023
At: Yokohama, Japan
1. Muhammad Ibrahim, Aleix Beneyto, Ivan Contreras, and Josep Vehi P hD. "Faults And Fault Tolerance In Automated Insulin Delivery Systems With An Emphasis On Human-In-The-Loop." IFAC-PapersOnLine 56, no. 2 (2023): 11503-11514.
DOI: https://doi.org/10.1016/j.ifacol.2023.10.441
DOI: https://doi.org/10.1016/j.ifacol.2023.10.441
2020
IEEE International Conference (Peer-reviewed)
International Conference on Advances in the Emerging Computing Technologies (AECT) 2019-2020
At: Al Madinah Al Munawwarah, Saudi Arabia, Saudi Arabia
2. Shafeeq, Muhammad, Muhammad Ibrahim, Zahid Ullah, Abdul Hafeez, and Tama Fouzder. "A Wearable Millimeter Wave MIMO Antenna Design For High Frequency Applications." In 2019 International Conference on Advances in the Emerging Computing Technologies (AECT), pp. 1-4. IEEE, 2020.
DOI: 10.1109/AECT47998.2020.9194218
DOI: 10.1109/AECT47998.2020.9194218
3. Rehman, Najib Ur, Omer Mujahid, Zahid Ullah, Abdul Hafeez, Tama Fouzder, and Muhammad Ibrahim. "Power Efficient FPGA-based TCAM Architecture by using Segmented Matchline Strategy." In 2019 International Conference on Advances in the Emerging Computing Technologies (AECT), pp. 1-4. IEEE, 2020.
DOI: 10.1109/AECT47998.2020.9194189
DOI: 10.1109/AECT47998.2020.9194189
2020
Title
"A Fast Mutation Detection System for Cancer Prevention on FPGA"
"A Fast Mutation Detection System for Cancer Prevention on FPGA"
Abstract
Fast variants detection in the genome sequences is the need of the hour. Thus, fast mutation detection system requires a fast computational speed (such as, parallel processing or parallel computation). Due to this, a content-addressable memory (CAM) is used in the proposed system; it is novel approach in the field of mutation detection. Especially, using logic-based CAM (BiCAM) helped us to performed exact matching, and the latency of the system is just 2 clock cycle. The proposed system is implemented on FPGA using the sequence of TP53 generated protein (p53). The p53 is also known as “the guardian of the genome”, therefore, mutation occurred in p53 plays a vital role in tumor creation. The p53 sequence is obtained from publicly available database of International Agency for Research on Cancer – TP53 database IARC.
The place and route simulation and implementation of the proposed system show that the system is power efficient (27 mW power consumption), 100% accurate (as BiCAM perform exact matching), and speedy as well (having pattern matching time equals to 0.18 µs). Thus, the proposed system is much efficient than the latest prior work. An addition to this, hamming distance calculating algorithm is also used to find out the exact number of nucleotides in the particular match-line which are mutated.
Fast variants detection in the genome sequences is the need of the hour. Thus, fast mutation detection system requires a fast computational speed (such as, parallel processing or parallel computation). Due to this, a content-addressable memory (CAM) is used in the proposed system; it is novel approach in the field of mutation detection. Especially, using logic-based CAM (BiCAM) helped us to performed exact matching, and the latency of the system is just 2 clock cycle. The proposed system is implemented on FPGA using the sequence of TP53 generated protein (p53). The p53 is also known as “the guardian of the genome”, therefore, mutation occurred in p53 plays a vital role in tumor creation. The p53 sequence is obtained from publicly available database of International Agency for Research on Cancer – TP53 database IARC.
The place and route simulation and implementation of the proposed system show that the system is power efficient (27 mW power consumption), 100% accurate (as BiCAM perform exact matching), and speedy as well (having pattern matching time equals to 0.18 µs). Thus, the proposed system is much efficient than the latest prior work. An addition to this, hamming distance calculating algorithm is also used to find out the exact number of nucleotides in the particular match-line which are mutated.
Supervised by
Dr. Azhar Qazi
Head of Department
Assistant Professor
Department of Electrical Engineering
CECOS University, Peshawar, Pakistan
Dr. Azhar Qazi
Head of Department
Assistant Professor
Department of Electrical Engineering
CECOS University, Peshawar, Pakistan
2017
Title
"Pattern Recognition Through BVL-CAM"
"Pattern Recognition Through BVL-CAM"
Abstract
Local Binary Patterns (LBP) is a type of visual descriptor that is used in feature classification. Local Binary Patterns are effectively used as a pattern recognition technique in pattern recognition systems. Many biometric recognition systems, i.e. Face recognition and retina recognition make use of LBP. Established pattern recognition systems based on LBP are deployed using conventional RAM. A Content addressable memory is a specific type of computer memory that can perform fast search operations in one clock cycle. Using CAM for pattern recognition effectively improves the performance of any pattern recognition system that previously used RAM. This research proposes an alternative way to design a Content Addressable Memory by using existing RAM resources of an FPGA device. We call this type of CAM, the Binary Validation Logic CAM, BVL-CAM. Our proposed method uses two FPGA RAMs and a control logic circuit to emulate the functionality of a CAM. Our proposed CAM based system design offers a lot of advantages over the established RAM based techniques, such as improved speed and low power consumption. We have used our proposed content addressable memory model to compute histograms for Local Binary Patterns (LBP). By using BVL-CAM for LBP computation we reduce the amount of time required to compute LBP histogram and hence improve the speed of the system.
Local Binary Patterns (LBP) is a type of visual descriptor that is used in feature classification. Local Binary Patterns are effectively used as a pattern recognition technique in pattern recognition systems. Many biometric recognition systems, i.e. Face recognition and retina recognition make use of LBP. Established pattern recognition systems based on LBP are deployed using conventional RAM. A Content addressable memory is a specific type of computer memory that can perform fast search operations in one clock cycle. Using CAM for pattern recognition effectively improves the performance of any pattern recognition system that previously used RAM. This research proposes an alternative way to design a Content Addressable Memory by using existing RAM resources of an FPGA device. We call this type of CAM, the Binary Validation Logic CAM, BVL-CAM. Our proposed method uses two FPGA RAMs and a control logic circuit to emulate the functionality of a CAM. Our proposed CAM based system design offers a lot of advantages over the established RAM based techniques, such as improved speed and low power consumption. We have used our proposed content addressable memory model to compute histograms for Local Binary Patterns (LBP). By using BVL-CAM for LBP computation we reduce the amount of time required to compute LBP histogram and hence improve the speed of the system.
Group Members
Muhammad Ibrahim
Asim Ullah
Muhammad Ibrahim
Asim Ullah
Supervised by
Dr. Omer Mujahid
PostDoc Fellow, MICELab, University of Girona, Spain
Dr. Omer Mujahid
PostDoc Fellow, MICELab, University of Girona, Spain