2024, Vol. 31
Early stages of diabetes can be effectively treated via diet, exercise, and constant blood sugar monitoring. This may also help to avoid the arrival of type Ⅱ diabetes[1]. Diabetes is manageable. If suitable lifestyle adjustments are achieved, it can even go into remission in certain circumstances. A patient who has had type Ⅱ diabetes for more than 10 years will have some sort of nerve damage[2]. Nerves connected to eyes and foot are mostly affected. A patient with diabetes is at risk of developing a foot ulcer. Insulin users, as well as people with diabetes related heart illness, eye and kidney are more possible to grow a foot ulcer[3]. Obesity, as well as the use of alcohol and tobacco, contributes to the growth of foot ulcers.
Diabetic Foot Ulcers (DFU) are caused by a number of causes, including an absence of sensation in the feet, reduced circulation, trauma, foot abnormalities and irritation, as well as the length of time that a person has had diabetes[4]. Neuropathy, a diminished or full lack of capacity to sense pain in the feet caused by nerve damage induced by rising blood glucose levels with time, can develop in patients with diabetes who have had it for a long period. Nerve injury is a common occurrence[5], 15% of diabetic patients are affected by DFU in their lifetime. 64 million people are affected by diabetics in India and 40000 amputations are done every year. Worldwide Statistical analysis of DFU amputation is given in Table 1, where 50% of diabetic patients are affected by ulcer in the UK in which 20% of the ulcer patients get amputated.100000 amputations are done every year worldwide. Additional risk factors, such as smoking, high blood glucose, drinking alcohol, and high cholesterol must be minimised in order to avoid and cure a diabetic foot ulcer[6]. One can lower the risks by wearing the correct shoes and socks. Podiatrist can assist diabetic patients in selecting the appropriate footwear. Due to inadequate glycaemic management and incorrect foot care, the prevalence of foot disorders is higher among low socioeconomic groups. The expense of treating diabetic patients' foot problems, both directly and indirectly, is substantial[7]. Preventive actions must be implemented immediately in order to lessen the expense burden on patients and society. According to the American Diabetes Association, people having diabetes should undergo a complete foot investigation once a year[8]. The majority of foot ailments can be avoided with proper foot care. It may take some time and effort to develop proper foot care routines, but self-care is critical[9]. Causes, diagnosis, factors and risk reduction techniques in concern with diabetic foot ulcer is shown in Fig. 1.
|
Table 1 Prevalence rate of diabetic foot ulcer |
|
Fig.1 Causes, diagnosis, factors and risk reduction techniques in concern with diabetic foot uicer[10] |
Furthermore, wearing improper footwear, such as chappals with a rubber puts the feet at risk of damage[11]. When shoes are worn, they tend to be pointed, which exposes the foot to more harm. According to statistical analysis of prevalence rate with infection and amputation rate of the U.S, U.K. and India, the prevalence is greater in Males. type Ⅱ diabetic patients prevalence rate is greater compared with the other types of diabetes.
If the blood sugar level is high and constant pressure is exerted on the ulcer, it may take longer for it to heal. The most efficient strategy to allow foot ulcers to heal is to maintain a diet that faciliates the achievement of glycaemic targets while also off-loading pressure on the feet. After an ulcer has healed, persistent preventive care can help a patient avoid a recurrence. To prevent foot at risk of ulceration, patients must be educated and monitored throughout their lives[12].
Patients should be aware that feet at high risk require careful use and consideration. The amount of activity should be kept to a bare minimum. Effective foot examinations, studies to identify foot ulcers, and proper care procedures, as well as preventive measures, would go a long way toward limb salvage and prevention. Various stages of diabetic foot are described in Fig. 2. and in Table 2
|
Fig.2 Stages of diabetic foot |
|
|
Table 2 Stages and descriptions |
We proposed a fuzzy based recognition model since fuzzy models are important in classification problems because they allow for a more flexible and nuanced approach to decision-making compared to traditional binary classification methods. Fuzzy models can also incorporate expert knowledge or domain-specific information into the model, which can improve the accuracy of the classification. Overall, fuzzy models can provide more accurate and reliable predictions in complex classification problems.
The rest of this article is structured as follows: Related work is discussed in Section 1. Section 2 describes our proposed system using multi layered fuzzy model. Section 3 presents the results and Section 4 winds up with the conclusion.
1 Related WorkResearchers have explored various techniques for predicting diabetes using different prediction models. These techniques include neural network models, genetic regression models, datamining regression techniques, and fuzzy expert systems. For example, some studies [13-15] used the SVM method to identify patterns in diabetes data and provided better treatment processes for diabetes in different age groups. Other studies have employed fuzzy C-means clustering algorithms to identify diabetes mellitus retinopathy characteristics, while others have developed five-layer fuzzy ontologies in diabetic data. Hybrid fuzzy logic and the artificial bee colony algorithm have been used for classification and the diagnosis of diabetes. In addition, some studies [16-18] have employed machine learning algorithms such as the Gini index-Gaussian fuzzy decision trees to predict early detection of diabetes. Collecting patient physiological information and converting it into vector input has also been used to identify Type 2 diabetes mellitus. The proposed techniques aimed to address the complexity of healthcare data, as the processed representation was dealing with high-dimensional data. The various attempts to add fuzzy-based applications to prevent early recognition of diabetics are a promising area of research [19-21]. Shakeel et al.[1] proposed a cloud-based framework for the diagnosis of diabetes mellitus using K-means clustering. The framework used a dataset of 768 instances with 8 features to train a K-means clustering algorithm. The authors achieved an accuracy of 78.64% for their model. Sheikhpour et al.[22] proposed an intelligent approach to diagnose diabetes using bi-level dimensionality reduction and classification algorithms. The proposed approach involves using genetic algorithms for feature selection and principal component analysis for dimensionality reduction. The reduced feature set is then used as input to various classification algorithms including artificial neural networks, k-nearest neighbors, and decision trees. The approach was evaluated on the Pima Indian diabetes dataset, and the results showed that the proposed approach outperformed other existing methods in terms of accuracy and sensitivity. Tuttolomondo et al.[23] explored the relationship between diabetic foot syndrome (DFS) and cardiovascular disease (CVD) in patients with diabetes. The authors suggested that DFS may serve as a marker for CVD, as both conditions are linked to chronic inflammation and endothelial dysfunction. They reviewed the evidence linking DFS to an increased risk of CVD, and proposed that early detection and management of DFS may help reduce the incidence of CVD in diabetic patients. The article also discussed potential mechanisms underlying the association between DFS and CVD, including shared risk factors such as hyperglycaemia and hypertension, as well as direct effects of foot ulcers and amputations on cardiovascular function. Patra[14] proposed an expert system for the automatic diagnosis of diabetes. The system is designed to take in various patient parameters such as age, gender, body mass index, blood pressure, and family history, and to provide a diagnosis of diabetes. Various components of the expert system such as the knowledge base, inference engine, and user interface were described in the paper. The system was tested on a dataset of 500 patient records and was able to diagnose diabetes with an accuracy of 91.6%. Jayashree et al.[4] proposed a fuzzy expert system to predict the risk of diabetic foot ulcers based on various input parameters. The proposed model uses a classification system to distinguish between each fuzzy framework and its parameters, and recommends necessary prevention, treatment, and medication based on the severity of the risk. Kamalakannan et al.[24] developed deep learning methods to achieve real-time localization of DFU. They also presented a case study that demonstrates the effectiveness of the proposed model in identifying the risk of diabetic foot ulcer in patients. Overall, the paper highlights the potential of fuzzy expert systems in the prediction and management of diabetic foot ulcer, which is a common and serious complication of diabetes. The referenced works collectively contribute to the advancement of diabetic foot ulcer detection and monitoring. Goyal et al.[21] emphasized the importance of mobile technology for real-time ulcer detection. Kaselimi et al.[25] provided a comprehensive review of non-invasive sensors and AI models in diabetic foot monitoring. Doulamis et al. [26] introduced a photonics-based device for non-invasive monitoring. Tzortzis et al.[27] focused on unsupervised monitoring techniques. Thotad et al.[28] employed deep learning for ulcer detection, and Kairys et al.[29] reviewed home-based monitoring, collectively offering diverse approaches to enhance the management of diabetic foot ulcers.
The limitation of the existing studies is the lack of detailed information of the various models used in these studies. They did not provide sufficient information on the characteristics of the dataset and its size, which can have an impact on the performance of the proposed approach. Another drawback is that the study only considers a single type of classifier (K-Nearest Neighbors), which may limit the generalizability of the results. To overcome this, the proposed work introduced a fuzzy based recognition model which can provide more accurate and reliable predictions in complex classification problems.
2 Proposed MethodThe proposed fuzzy model utilizes various independent fuzzy frameworks consisting of inference parameters and fuzzy expert systems to predict the risk of foot ulcers in type-Ⅱ diabetic patients. Our architecture is split into 2 levels. The primary level consists of agents which work on inference values. The second level consists of a multi-layered architecture comprising of 4 FFs (Fuzzy Frameworks) which work along a knowledge domain to predict the danger of DFU and measure its severity across 5 different levels.
The model works on a rule-based system, which takes into consideration various input parameters such as condition of feet, BMI, age, duration of diabetes, co-morbidity, sensory loss to vibration and then categorizes into subgroups. These values are then sent to their respective frameworks for further calibration and prediction by the Inference Parameter Classifier (IPC). The dataset that has been used for this study is taken from Kaggle[30], which is an online dataset database. This dataset includes several attributes through which we have predicted whether the patient can get diabetes or not. All the instances of the dataset are women and at least 21 years old. The dataset is comprised of 768 patients, out of which 268 samples are identified as diabetic while 500 samples are identified as non-diabetic. Then the data set was used for testing and training. In the proposed layer 1 model, out of 500 classes, 250 classes are used for training and 250 classes are used for testing. In layer 2, 150 is used for training and 118 is used for testing out of 268.
2.1 Multi Layered Fuzzy ModelThis section elaborates on the different components present in the multi layered fuzzy model and the architecture diagram.
2.1.1 Regulating layerThe regulating layer consists of four agents, such as foot-related agent, physical-related agent, patient information-based agent, and patient character-related agent. Foot based agent is correlated with elements like sensory loss to vibration, type of footwear, foot deformity, etc. The second agent, the physical related agent checks on parameters like BMI, systolic blood pressure, co-morbidity, etc. The third agent functions by taking into account patient information like alcohol consumption, obesity, tobacco use, etc. The fourth agent depends on the attitude and knowledge of the patient about diabetes. The agent's job is to convert the values into fuzzy value and also to measure according to the fuzzy frameworks which work on them.
2.1.2 Fuzzy systemThe first-level fuzzy frameworks are given names as FF1, FF2, FF3, FF4. Each one of these frameworks works based on its rule-based system, which is used to generate values that then help us to predict the risk and severities. The rule-based system is derived from a knowledge base taken from various fuzzy expert systems. The entire process is sequential and the outputs generated from one framework are used in another. This second level formed in the architecture helps to predict the risk and severities and helps us categorize them into different levels.
2.1.3 Inference parameter classifierTo distinguish and categorize each of the parameters, specific classifiers are needed[31]. With the algorithms mentioned and the intensive study done on various classifiers, we came up with an Inference Parameter Classifier (IPC). IPC is responsible for taking out necessary attributes from the given dataset and assigning them to their relevant agents. To improve the accuracy of the IPC significantly, we use data mining to filter out all the attributes. The filtration process takes place among 4 different agents namely: food-related agents, physical related agents, patient information based agents, patient personality agents. The dataset of all the parameters used is given in Table 3.
|
|
Table 3 Dataset description |
2.2 Architecture Diagram
The workflow of fuzzy system based on various agents to give out the risk measures and necessary preventive measures, treatment and medications is depicted in Fig. 3.
|
Fig.3 Architecture diagram |
Based on the defined rules and input given in the fuzzy system [32-33], the severity of DFU is predicted. The priorities of the rules may differ based upon the importance of the factors that decide the risk of foot ulcers[34]. Here mamdani fuzzy logic inference system was used because of its easy-to-understand nature and wide acceptance. De-fuzzification is done by using centroid method. Various parameters discussed in Table 4 are selected using MATLAB FIS editor. "AND" method represents minimum (min), "OR" method represents maximum (max), implication method represents minimum (min) and aggregation method represents maximum(max).
|
|
Table 4 Parameters used in framework |
3 Results of Fuzzy Framework 3.1 Fuzzy Rules for Layer 1
Layer 1 consists of 4 fuzzy frameworks comprising of 20 inference parameters. Every fuzzy framework has its rule base for its respective parameters, those rules were developed by people who have expertise in the medical field[35]. These rules are written by comparing different inference parameters within the fuzzy frameworks. The results from layer 1 are used to generate results of layer 2. The rules to help us analyze the risk of the DFU for all the various inference parameters can be found here.
1) If (FP11 is present), (FP14 is present) and (FP13 is true), (Risk_of_DFU is very severe).
2) If (FP12 is present), (FP14 is true) and (FP11 is present), (Risk_of_DFU is high).
3) If (FP23 is high), (FP26 is yes), (BMI is high) and (Diastolic is high), (Risk_of_DFU is high).
4) If (FP26 Yes), (FP21 is high), (FP23 is high) and (Diastolic is high), (Risk_of_DFU is very severe).
5) If (FP36 is urban), (Attitude is unfavorable) and (Knowledge about diabetes is poor), (Risk_of_DFU is medium).
6) If (FP35 is non-farmer), (FP36 is rural) and (FP31 is yes), (Risk_of_DFU is high).
7) If (FP41 is non-favorable), (FP42 is unfavorable) and (FP43 is poor), (Risk_of_DFU is medium).
8) If (FP41 is favorable), (FP42 is poor) and (FP43 is poor), (Risk_of_DFU is medium)
3.2 Fuzzy Rules for Layer 2Each fuzzy framework has its own rule base, these rules were developed by people who have expertise in the medical field and are written by comparing different fuzzy frameworks and using different attributes from the fuzzy frameworks. The rules for all the fuzzy frameworks that contribute to the analysis of the risk of the DFU can be found here.
1) If (FF1 is high), (FF2 is medium) and (FF4 is high), Risk of DFU is severe.
2) If (FF2 is low), (FF3 is medium) and (FF4 is medium), Risk of DFU is moderate.
3) If (FF1 is high), (FF3 is medium) and (FF4 is high), Risk of DFU is severe.
4) If (FF1 is high), (FF2 is medium) and (FF3 is low), Risk of DFU is moderate.
4 Experimental EvaluationsThe results of the second layer were generated by comparing 4 fuzzy frameworks mentioned above. From the given results, it can be seen that all the 4 frameworks show a different trend towards the risk of DFU. In FF1 and FF4, as the values go higher the risk of DFU becomes more severe, whereas it is quite the opposite with FF2 and FF3. Depending on the trend shown by the frameworks, we categorised the risks into severe and moderate. Based on the classification necessary medications and drugs have been recommended. The results for FF1, FF2, FF3 and second layer inference are given in Tables 5-7 respectively.
|
|
Table 5 Results for the first fuzzy framework |
|
|
Table 6 Results for the second fuzzy framework |
|
|
Table 7 Results for the third fuzzy framework |
A detailed analysis of a fuzzy model proposed for predicting diabetes is provided in this section. The study used a diabetes dataset sourced from the Kaggle online database and implemented on MATLAB R2021a (version 10.0). The dataset was normalized, with numerical values ranging from 0 to 1. Table 8 shows the confusion matrix. The proposed fuzzy rules were then used to classify the dataset, and a confusion matrix was constructed as in Table 9 to evaluate the classifiers. The confusion matrix showed the number of true negatives (TN), false positives (FP), false negatives (FN), and true positives (TP) for both classifiers, with 52% of the data used for training and 48% for testing.
|
|
Table 8 Confusion matrix |
|
|
Table 9 Accuracy analysis |
Layer 1 model predicted "yes" 121 times and "no" 247 times out of a total of 368 predictions, with 113 TP, 5 FN, 8 FP, and 242 TN. Layer 2 predicted "yes" 136 times and "no" 232 times, with 106 TP, 12 FN, 5 FP, and 245 TN. In reality, there were 118 diabetic samples and 250 non-diabetic samples. The proposed classifiers' accuracies were compared with other fuzzy classification methods, and the results showed that they outperformed the state-of-the-art techniques in classification accuracy.
The classifiers' accuracy, precision, recall, and F-measure were evaluated, and the results showed that both classifiers performed well. Classifier 1 achieved scores of 96.47%, 95.76%, 93.39%, and 94.56% for accuracy, recall, precision, and F-measure, respectively, while classifier 2 scored 95.38%, 89.83%, 95.50%, and 92.58% for the same parameters. Overall, the proposed fuzzy classifiers showed great potential for detecting diabetes.
5 ConclusionsThe proposed fuzzy model predicts the risk of foot ulcers in type-Ⅱ diabetic patients using various independent fuzzy frameworks consisting of inference parameters and layers of fuzzy expert systems. The model uses a classification system to distinguish between each fuzzy framework and its parameters. Based on the severity levels necessary prevention, treatment, and medication are recommended. When the risk of DFU is high, a consultation with a podiatrist for proper-fitting shoes is recommended as a preventive measure. As a treatment, removing the pressure from the area, getting rid of dead skin and tissue is considered effective and necessary. When the risk of DFU is low or moderate, washing your feet and keeping them dry and moisturised is recommended as an effective preventive measure. As a treatment, using a wound dressing, cleaning the wound on a daily basis is recommended. The proposed model is compared with other related models used in the work. From the analysis, we confirm that the proposed fuzzy classifiers such as layer 1 and layer 2 achieved higher accuracy than the others.
Combining the results of all the fuzzy frameworks derived from its constituent parameters, a risk-specific medication is recommended. A surgeon can assist you in reducing the amount of pressure around your ulcer by removing foot abnormalities when the risk is severe.
Foot ulcers are curable if diagnosed early. If a diabetic patient gets a sore on foot, he/she has to see a doctor straight away, as the longer he/she waits, the greater the risk of infection. Amputations may be required if infections are not managed. Until the ulcer heals, stay off your feet and follow your course of treatment. It could take multiple weeks for diabetic foot sores to heal.
| [1] |
Shakeel P M, Baskar S, Dhulipala V S, et al. Cloud based framework for diagnosis of diabetes mellitus using K-means clustering. Health Information Science and Systems, 2018, 6: 1-7. DOI:10.1007/s13755-018-0054-0 (  0) |
| [2] |
Joshi S, Borse M. Detection and prediction of diabetes mellitus using back-propagation neural network. Proceedings of the International Conference on Micro-electronics and Telecommunication Engineering (ICMETE). New York: IEEE Xplore, 2016: 110-113. DOI: 10.1109/ICMETE.2016.11.
( 0) |
| [3] |
Osman A H, Aljahdali H M. Diabetes disease diagnosis method based on feature extraction using K-SVM. International Journal of Advanced Computer Science and Applications, 2017, 8(1): 236-244. DOI:10.14569/IJACSA.2017.080130 ( 0) |
| [4] |
Jayashree J, Chaudhari N, Iyengar N C S. Promethean fuzzy model to predict diabetic foot ulcer. International Journal of Hybrid Information Technology, 2016, 9: 181-198. DOI:10.14257/ijhit.2016.9.11.16 ( 0) |
| [5] |
Purnami S W, Zain J M, Embong A. A new expert system for diabetes disease diagnosis using modified spline smooth support vector machine. Proceedings of the International Conference on Computational Science and its Applications. Heidelberg: Springer, 2010: 83-92.
( 0) |
| [6] |
Jayashree J, Sriman N C. A review on mobile based intelligent systems for homecare monitoring of diabetic mellitus foot ulcer. International Journal of Smart Home, 2016, 10: 201-212. DOI:10.14257/IJSH.2016.10.8.24 ( 0) |
| [7] |
Bhandari V, Kumar R. Comparative analysis of fuzzy expert systems for diabetic diagnosis. International Journal of Computer Applications, 2015, 132: 8-14. DOI:10.5120/ijca2015907424 ( 0) |
| [8] |
Nnamoko N A, Arshad F, England D, et al. Fuzzy expert system for type 2 diabetes mellitus (T2DM) management using dual inference mechanism. Proceedings of the 2013 AAAI Spring Symposium Series. Berlin: Springer, 2013.
( 0) |
| [9] |
Luukka P. Feature selection using fuzzy entropy measures with similarity classifier. Expert Systems with Applications, 2011, 38(4): 4600-4607. DOI:10.1016/j.eswa.2010.09.133 ( 0) |
| [10] |
Diabetic Foot Ulcers-Symptoms, Causes, And Risk Factors. Available online from https://www.pristyncare.com/blog/diabetic-foot-ulcers-symptoms-causes-risk-factors-pc0441/
( 0) |
| [11] |
Choubey D K, Paul S. GA_MLP NN: a hybrid intelligent system for diabetes disease diagnosis. International Journal of Intelligent Systems and Applications, 2016, 8: 49-59. DOI:10.5815/IJISA.2016.01.06 ( 0) |
| [12] |
Feroz Khan A B, Kamalakannan K, Syed Siraj Ahmed N. Integrating machine learning and stochastic pattern analysis for the forecasting of time-series data. SN Computer Science, 2023, 4: Article Number: 484. DOI:10.1007/s42979-023-01981-0 ( 0) |
| [13] |
Choubey D K, Paul S, Bhattacharjee J. Soft computing approaches for diabetes disease diagnosis: a survey. International Journal of Applied Engineering Research, 2014, 9(21): 11715-11726. ( 0) |
| [14] |
Patra P S K. Automatic diagnosis of diabetes by expert system. International Journal of Computer Science Isues, 2012, 9(2): 299-304. ( 0) |
| [15] |
Jayashree J, Kumar S A. Linear discriminant analysis based genetic algorithm with generalized regression neural network-a hybrid expert system for diagnosis of diabetes. Programming and Computer Software, 2018, 44: 417-427. DOI:10.1134/S0361768818060063 ( 0) |
| [16] |
Díaz-Redondo A, Giráldez-García C, Carrillo L, et al. Modifiable risk factors associated with prediabetes in men and women: A cross-sectional analysis of the cohort study in primary health care on the evolution of patients with prediabetes (PREDAPS-Study). BMC Family Practice, 2015, 16: Article number: 5. DOI:10.1186/s12875-014-0216-3 ( 0) |
| [17] |
Jayashree J, Kumar S A. Hybrid swarm intelligent redundancy relevance (RR) with convolution trained compositional pattern neural network expert system for diagnosis of diabetes. Health and Technology, 2020, 10: 197-206. DOI:10.1007/s12553-018-00291-3 ( 0) |
| [18] |
Li Y, Li H, Yao H. Analysis and study of diabetes follow-up data using a data-mining-based approach in new urban area of Urumqi, Xinjiang, China, 2016-2017. Computational and Mathematical Methods in Medicine, 2018, 2018: Article ID 7207151. DOI:10.1155/2018/7207151 ( 0) |
| [19] |
Vaishali R, Sasikala R, Ramasubbareddy S, et al. Genetic algorithm based feature selection and MOE Fuzzy classification algorithm on Pima Indians diabetes dataset. Proceedings of the 2017 International Conference on Computing Networking and Informatics (ICCNI), Lagos, Nigeria, 2017: 1-5. DOI: 10.1109/ICCNI.2017.8123815.
( 0) |
| [20] |
Bhuvaneswari G, Manikandan G. A novel machine learning framework for diagnosing the type 2 diabetics using temporal fuzzy ant miner decision tree classifier with temporal weighted genetic algorithm. Computers, 2018, 100: 759-772. DOI:10.1007/s00607-018-0599-4 ( 0) |
| [21] |
Goyal M, Reeves N D, Rajbhandari S, et al. Robust methods for real-time diabetic foot ulcer detection and localization on mobile devices. IEEE Journal of Biomedical and Health Informatics, 2019, 23(4): 1730-1741. DOI:10.1109/JBHI.2018.2868656 ( 0) |
| [22] |
Sheikhpour R, Sarram M A. Diagnosis of diabetes using an intelligent approach based on bi-level dimensionality reduction and classification algorithms. Iranian Journal of Diabetes and Obesity, 2014, 6: 74-84. ( 0) |
| [23] |
Tuttolomondo A, Maida C, Pinto A. Diabetic foot syndrome as a possible cardiovascular marker in diabetic patients. Journal of Diabetes Research, 2015, 2015: 268390. DOI:10.1155/2015/268390 ( 0) |
| [24] |
Kamalakannan K, Anandharaj G, Gunavathie M A. Performance analysis of attributes selection and discretization of Parkinson's disease dataset using machine learning techniques: A comprehensive approach. International Journal of System Assurance Engineering and Management, 2023, 14: 1523-1529. DOI:10.1007/s13198-023-01960-x ( 0) |
| [25] |
Kaselimi M, Protopapadakis E, Doulamis A, et al. A review of non-invasive sensors and artificial intelligence models for diabetic foot monitoring. Frontiers in Physiology, 2022, 13: 924546. DOI:10.3389/fphys.2022.924546 ( 0) |
| [26] |
Doulamis A, Doulamis N, Angeli A, et al. A non-invasive photonics-based device for monitoring of diabetic foot ulcers: architectural/sensorial components & technical specifications. Inventions, 2021, 6(2): 27. DOI:10.3390/inventions6020027 ( 0) |
| [27] |
Tzortzis I N, Davradou A, Protopapadakis E, et al. Unsupervised diabetic foot monitoring techniques. Proceedings of the 15th International Conference on PErvasive Technologies Related to Assistive Environments. 2022: 608-614. DOI: 10.1145/3529190.3534723.
( 0) |
| [28] |
Thotad P N, Bharamagoudar G R, Anami B S. Diabetic foot ulcer detection using deep learning approaches. Sensors International, 2023, 4: 100210. DOI:10.1016/j.sintl.2022.100210 ( 0) |
| [29] |
Kairys A, Pauliukiene R, Raudonis V, et al. Towards home-based diabetic foot ulcer monitoring: A systematic review. Sensors, 2023, 23(7): 3618. DOI:10.3390/s23073618 ( 0) |
| [30] |
Kaggle. Pima indians diabetes database. https://www.kaggle.com/datasets/uciml/pima-indians-diabetes-database. [accessed on 27 November 2022].
( 0) |
| [31] |
Dania Sadaf, Javeria Amin, Muhammad Sharif, et al. Detection of diabetic foot ulcer using machine/deep learning. Advances in Deep Learning for Medical Image Analysis, 2022, 101-123. ( 0) |
| [32] |
Jain V, Raheja S. Improving the prediction rate of diabetes using fuzzy expert system. International Journal of Information Technology and Computer Science, 2015, 10: 84-91. DOI:10.5815/IJITCS.2015.10.10 ( 0) |
| [33] |
Kalpana M, Kumar A S. Fuzzy expert system for diabetes using fuzzy verdict mechanism. Int. J. Advanced Networking and Applications, 2011, 3(2): 1128-1134. ( 0) |
| [34] |
Jayashree J, Kumar S A. Evolutionary correlated gravitational search algorithm (ECGS) with genetic optimized Hopfield neural network (GHNN)-A hybrid expert system for diagnosis of diabetes. Measurement, 2019, 145: 551-558. DOI:10.1016/j.measurement.2018.12.08 ( 0) |
| [35] |
Peppa M, Stavroulakis P, Raptis S A. Advanced glycoxidation products and impaired diabetic wound healing. Wound Repair and Regeneration, 2009, 17: 461-472. DOI:10.1111/j.1524-475X.2009.00518.x ( 0) |
| [36] |
Chen T, Shang C, Su P, et al. Effective diagnosis of diabetes with a decision tree-initialised neuro-fuzzy approach. Singapore: Springer, 2018, 840: 227-239. ( 0) |
| [37] |
Cheruku R, Edla D R, Kuppili V, et al. RST-BatMiner: A fuzzy rule miner integrating rough set feature selection and Bat optimization for detection of diabetes disease. Applied Soft Computing, 2018, 67: 764-780. DOI:10.1016/j.asoc.2017.06.032 ( 0) |
| [38] |
Singh A, Dhillon A, Kumar N, et al. eDiaPredict: an ensemble-based framework for diabetes prediction. ACM Transactions on Multimedia Computing, Communications, and Applications, 2021, 17(Article No.: 66): 1-26. DOI:10.1145/3415155 ( 0) |