CoChat
Sign Up
Cited by 366Privacy-Preserving Technologies in DataInternet of Things and AIIoT and Edge/Fog Computing

Federated Learning for Privacy Preservation in Smart Healthcare Systems: A Comprehensive Survey

Open access
Ali, Mansoor, Naeem, Faisal, Tariq, Muhammad, Kaddoum, Georges·2023·IEEE Journal of Biomedical and Health Informatics
DOI 10.1109/jbhi.2022.3181823
Open this paper in CoChat — free accountOpen PDF

Abstract

Recent advances in electronic devices and communication infrastructure have revolutionized the traditional healthcare system into a smart healthcare system by using internet of medical things (IoMT) devices. However, due to the centralized training approach of artificial intelligence (AI), mobile and wearable IoMT devices raise privacy issues concerning the information communicated between hospitals and end-users. The information conveyed by the IoMT devices is highly confidential and can be exposed to adversaries. In this regard, federated learning (FL), a distributive AI paradigm, has opened up new opportunities for privacy preservation in IoMT without accessing the confidential data of the participants. Further, FL provides privacy to end-users as only gradients are shared during training. For these specific properties of FL, in this paper, we present privacy-related issues in IoMT. Afterwards, we present the role of FL in IoMT networks for privacy preservation and introduce some advanced FL architectures by incorporating deep reinforcement learning (DRL), digital twin, and generative adversarial networks (GANs) for detecting privacy threats. Moreover, we present some practical opportunities for FL in IoMT. In the end, we conclude this survey by discussing open research issues and challenges while using FL in future smart healthcare systems.

AI snapshot

Written from the paper’s full text.

TL;DR

This survey reviews privacy and security issues in IoMT and how federated learning (FL) can preserve privacy by training models collaboratively without accessing local data. It also surveys advanced FL architectures using DRL, digital twin, and GANs, plus applications and open research challenges for smart healthcare.

Key findings

  • FL enables distributed model training by averaging local model updates from multiple IoMT networks without accessing local data; only gradients are shared, which can mitigate disclosure of user preferences and sensitive patient information.
  • The survey identifies IoMT privacy threats including information leakage, poisoning attacks, Byzantine attacks, privacy data leakage attacks, and inference-based attacks, and reviews defenses such as Paillier homomorphic encryption, Shamir’s threshold secret sharing, GANs, blockchain, differential privacy, and homomorphic encryption.
  • Advanced FL designs for IoMT incorporate DRL for incentive allocation, digital twins for anomaly detection and secure remote patient monitoring, and GANs for detecting malicious users or generating privacy-preserving raw image datasets.
  • Applications discussed include electronic health record management, medical image processing, and COVID-19 detection; the survey reports that one cited FL-based image-processing algorithm provides 97% accuracy on cancer datasets and outperforms non-FL-based schemes.

Methods

  • Comprehensive survey design: the paper reviews privacy issues in IoMT, the role and architecture of FL in IoMT, advanced FL architectures, practical applications, and open research directions.
  • Data source is published literature; the paper presents comparison tables of existing FL healthcare surveys and advanced FL architectures for IoMT privacy preservation.
  • Main analysis is qualitative and taxonomic synthesis; no primary dataset, experiment, or statistical analysis is conducted by the survey itself.

Limitations

  • The paper states in Table II that several reviewed FL approaches need numerical methods for performance evaluation, more robust and lightweight encryption, solutions to FL convergence issues, incentive rewards to encourage node participation, and more robust digital-twin models for unseen privacy threats.
  • The paper also notes that some reviewed incentive models assume independent and identically distributed (IID) data.
  • Evident from the text: as a survey, it provides no new empirical validation, and it states that universal standards/protocols for evaluating FL in IoMT are still lacking.

Model-written summary. Check it against the paper before relying on it.

Integrity signals

Registry

Crossref · 10.1109/jbhi.2022.3181823

Cites

Statistics

Nothing to recompute.

Claims the paper makes

0 of 55 cited sources checked · 0 supported · 0 flagged

  • Findingown data

    Increasing privacy reduces the accuracy of the training algorithm.

    “It has been observed from the analysis of the results that by increasing privacy, the accuracy of the training algorithm is reduced.”
    Quote in the paperSubstantial claim0 of 1 sources checked
    • cites [65] source

      Not checked in the free run.

  • Premisereasoning

    The centralized AI framework for future adaptive healthcare systems may not be practical because health traffic will be distributed over heterogeneous and large-scale healthcare systems.

    “Moreover, the centralized AI framework for future adaptive healthcare systems may not be practical because health traffic will be distributed over heterogeneous and large-scale healthcare systems.”
    Quote in the paperSubstantial claim
  • Premisecitation

    Federated learning has shown promising solutions for providing privacy protection in smart-healthcare networks.

    “In this context, the emerging concept of federated learning (FL) has shown promising solutions for providing privacy protection in smart-healthcare networks [5], [6].”
    Quote in the paperSubstantial claim0 of 2 sources checked
    • cites [5] Yuan, 2020, A federated learning framework for healthcare IoT devices

      Not checked in the free run.

    • cites [6] source

      Not checked in the free run.

  • Premisecitation

    Federated learning improves training performance of health by collecting large datasets and computation resources from local IoMT devices, which might not be possible with centralized AI.

    “Moreover, the FL approach improves the training performance of the health by collecting large datasets and computation resources from local IoMT devices, which might not be possible in the case of using the centralized AI approach [6].”
    Quote in the paperNot a substantial claim0 of 1 sources checked
    • cites [6] source

      Not checked in the free run.

  • Premisecitation

    Given the growth of IoMT devices and large volumes of health traffic in modern smart health care systems, the centralized architecture results in scalability issues.

    “Given the growth of IoMT devices and large volumes of health traffic in modern smart health care systems, the centralized architecture results in scalability issues [3].”
    Quote in the paperSubstantial claim0 of 1 sources checked
    • cites [3] source

      Not checked in the free run.

  • Classified as probably not a substantial claim. 26
  • May combine two claims. 24

  • Classified as probably not a substantial claim. May combine two claims.

    Interpretationreasoning

    Security and privacy issues pose a significant threat to modern healthcare systems and need to be addressed for effective operation of IoMT systems.

    “Henceforth, security and privacy issues pose a significant threat to modern healthcare systems and need to be addressed for the effective operation of IoMT systems.”
    Quote in the paperSubstantial claim
    • Rests on an assumption: Electronic healthcare systems are considered an easy target for attackers due to poor defense systems for information security.

  • Classified as probably not a substantial claim. May combine two claims.

    Premisecitation

    Addressing security issues in IoT has been primarily focused, but more attention needs to be paid to protection of end-user privacy in modern electronic health care systems.

    “It has been observed that addressing the security issues was primarily focused [11]; however, more attention needs to be paid to the protection of end-user privacy in modern electronic health care systems [12], [13].”
    Quote in the paperSubstantial claim0 of 3 sources checked
    • cites [11] source

      Not checked in the free run.

    • cites [12] Pham, 2021, Fusion of federated learning and industrial Internet of Things: A survey

      Not checked in the free run.

    • cites [13] source

      Not checked in the free run.

  • Probably not a substantial claim: may not pass S2 (checkable).

    Interpretationreasoning

    A distributed AI-based framework is needed for successful deployment of intelligent healthcare systems to enable privacy-preserving and scalable healthcare applications.

    “As a result, for the successful deployment of intelligent healthcare systems, a distributed AI-based framework is needed for enabling privacy-preserving and scalable healthcare applications.”
    Quote in the paperNot a substantial claim

  • Probably not a substantial claim: may not pass S3 (load-bearing).

    Premisecitation

    IoMT can significantly improve the efficiency of the healthcare system by ensuring the safety of patients.

    “Through these advanced countermeasures, it has been observed that IoMT can significantly improve the efficiency of the healthcare system by ensuring the safety of patients [14], [15].”
    Quote in the paperNot a substantial claim0 of 2 sources checked
    • cites [14] source

      Not checked in the free run.

    • cites [15] source

      Not checked in the free run.

  • May combine two claims.

    Premisecitation

    Reliance on centralized servers for data learning makes them vulnerable to security threats and poses a significant threat to patient safety and privacy.

    “Furthermore, the reliance on centralized servers for data learning makes them vulnerable and exposed to various types of security threats and thus, poses a significant threat to patient's safety and privacy [4].”
    Quote in the paperSubstantial claim0 of 1 sources checked
    • cites [4] source

      Not checked in the free run.

  • Classified as probably not a substantial claim.

    Premisenone given

    Federated learning is a type of distributed AI that enables distributed training of models by averaging local model updates from multiple IoMT networks without accessing local data.

    “FL is a type of distributed AI that enables the models' distributed training by averaging the local model updates from multiple IoMT networks without accessing the local data.”
    Quote in the paperSubstantial claim

  • Probably not a substantial claim: may not pass S3 (load-bearing).

    Premisecitation

    The incentive mechanism for federated learning can be classified into three categories: incentive allocation based on end-device data, device participation, and resources available at end devices.

    “In a broad sense, the incentive mechanism for FL can be classified into three categories: incentive allocation based on end-device data, device participation, and resources available at end devices [66].”
    Quote in the paperNot a substantial claim0 of 1 sources checked
    • cites [66] source

      Not checked in the free run.

  • May combine two claims.

    Premisecitation

    A third-party platform used for data processing and storage can access the data and may infer private information.

    “Moreover, a thirdparty platform is used for data processing and storage, which can access the data and may infer private information [46].”
    Quote in the paperSubstantial claim0 of 1 sources checked
    • cites [46] source

      Not checked in the free run.

  • May combine two claims.

    Interpretationreasoning

    The adoption of federated learning can resolve many privacy-concentrated issues in IoMT, but federated learning does have some privacy problems.

    “The adoption of FL can resolve many privacy-concentrated issues in IoMT, but FL does have some privacy problems.”
    Quote in the paperSubstantial claim
    • Rests on an assumption: If an adversary gets access to the central server, user information can be extracted.

  • Probably not a substantial claim: may not pass S3 (load-bearing).

    Premisecitation

    The Stackelberg game was adopted to model dynamic variation in end devices computation resources and stable connection between IoMT devices and central aggregator in hospital.

    “For instance, the Stackelberg game was adopted to model the dynamic variation in end devices computation resources and stable connection between IoMT devices and central aggregator in hospital [67].”
    Quote in the paperNot a substantial claim0 of 1 sources checked
    • cites [67] source

      Not checked in the free run.

    • Rests on an assumption: In vanilla federated learning algorithms, all IoMT devices must participate in training and share their aggregation model, but in most cases this does not occur.
    • Rests on an assumption: Due to limited computation capability of IoMT devices and privacy and trustworthiness of third-party platforms, IoMT devices do not show willingness to contribute to federated learning approaches.

  • May combine two claims.

    Premisecitation

    Malicious events that compromise privacy of PII could include tampering with the connection between smart devices in the home or access to private information via wearable health monitoring devices.

    “Malicious events that compromise the privacy of PII could include tampering with the connection between smart devices in the home or access to private information via wearable health monitoring devices [20], [21].”
    Quote in the paperNot a substantial claim0 of 2 sources checked
    • cites [20] source

      Not checked in the free run.

    • cites [21] Rannenberg, 2018, Enabling identity management and privacy in a global context: ISO/IEc standardization in JTC 1/SC 27/WG 5

      Not checked in the free run.

  • May combine two claims.

    Interpretationreasoning

    These actions adopted by federated learning enable efficient bandwidth utilization and avoid network congestion in massive IoMT networks.

    “These actions adopted by FL enable efficient bandwidth utilization and avoid network congestion in massive IoMT networks.”
    Quote in the paperNot a substantial claim

  • May combine two claims.

    Premisenone given

    In traditional ML approaches for healthcare data analysis, data is stored on a central cloud and is vulnerable to adversary attacks.

    “For instance, in traditional ML approaches for healthcare data analysis, the data is stored on a central cloud and is vulnerable to adversary attacks.”
    Quote in the paperNot a substantial claim
    • Rests on an assumption: Electronic healthcare systems are considered an easy target for attackers due to poor defense systems for information security.

  • May combine two claims.

    Premisecitation

    In a real-world healthcare system, data gathered from a single medical lab is not enough, and an ML model based on that dataset will not give promising results.

    “In a real-world healthcare system, the data gathered from a single medical lab is not enough, and the ML model based on that dataset will not give promising results [47].”
    Quote in the paperSubstantial claim0 of 1 sources checked
    • cites [47] source

      Not checked in the free run.

    • Rests on an assumption: Insufficient data makes it impossible to train an accurate model because input data is imbalanced and lacks enough feature information.

  • May combine two claims.

    Interpretationreasoning

    Adopting federated learning algorithms in IoMT increases privacy and reduces information leakage scenarios.

    “This increases privacy and reduces information leakage scenarios.”
    Quote in the paperSubstantial claim
    • Rests on an assumption: Constraints associated with individual IoT devices, such as power and resources, limit federated learning operation.

  • Probably not a substantial claim: may not pass S3 (load-bearing).

    Premisecitation

    By adopting k-anonymity, privacy protection levels can be controlled and quantified.

    “Nevertheless, by adopting k-anonymity, the privacy protection levels can be controlled and quantified, which is the main advantage of the anonymity algorithm [28].”
    Quote in the paperNot a substantial claim0 of 1 sources checked
    • cites [28] source

      Not checked in the free run.

  • May combine two claims.

    Premisecitation

    ML-based privacy preservation is mostly used due to its efficiency, and federated learning is the more preferable technique.

    “Nowadays, ML-based privacy preservation is mostly used due to its efficiency [34], and the more preferable technique is federated learning (FL).”
    Quote in the paperNot a substantial claim0 of 1 sources checked
    • cites [34] source

      Not checked in the free run.

  • Probably not a substantial claim: may not pass S3 (load-bearing). May combine two claims.

    Premisecitation

    Google initially used federated learning to combat privacy concerns with traditional ML, as federated learning works collaboratively on end devices.

    “Initially, Google used FL as a means to combat the privacy concerns with traditional ML, as FL works collaboratively on end devices [35].”
    Quote in the paperNot a substantial claim0 of 1 sources checked
    • cites [35] source

      Not checked in the free run.

  • Probably not a substantial claim: may not pass S3 (load-bearing).

    Premisecitation

    ML-based intrusion detection algorithms can be used for identification of adversary actions.

    “Moreover, it has been observed that ML-based intrusion detection algorithms can also be used for the identification of adversary actions [32], [33].”
    Quote in the paperNot a substantial claim0 of 2 sources checked
    • cites [32] source

      Not checked in the free run.

    • cites [33] source

      Not checked in the free run.

  • Probably not a substantial claim: may not pass S3 (load-bearing).

    Premisecitation

    Privacy assurance can be achieved by using Blockchain and authentication algorithms.

    “For that purpose, privacy assurance can be achieved by using Blockchain, and authentication algorithms [30], [31].”
    Quote in the paperNot a substantial claim0 of 2 sources checked
    • cites [30] source

      Not checked in the free run.

    • cites [31] source

      Not checked in the free run.

  • May combine two claims.

    Premisecitation

    While using FL, each hospital trains their model based on locally available X-ray images and only shares gradients with a global model for computation, hence ensuring privacy preservation.

    “While using FL, each hospital trains their model based on locally available X-ray images and only shares gradients with a global model for computation, hence ensuring privacy preservation [78].”
    Quote in the paperSubstantial claim0 of 1 sources checked
    • cites [78] source

      Not checked in the free run.

  • May combine two claims.

    Interpretationreasoning

    Removing patient names from health records is not sufficient for security because hospitals and agencies collaborate and commonly share a single central server where patient data is stored.

    “For security purposes, the patient name is removed from health records, but this solution is not sufficient as hospitals and agencies work in collaboration to find the solution of new diseases, and they commonly share a single central server where patient data is stored.”
    Quote in the paperSubstantial claim

  • May combine two claims.

    Premisecitation

    The algorithm provides 97% accuracy on cancer datasets and outperforms other non-FL-based schemes.

    “It has been observed using experimental analysis that the algorithm provides 97% accuracy on cancer datasets and outperforms other non-FL-based schemes.”
    Quote in the paperNot a substantial claim0 of 1 sources checked
    • cites [73] source

      Not checked in the free run.

  • Probably not a substantial claim: may not pass S3 (load-bearing). May combine two claims.

    Premisecitation

    Authors in [69] find a solution by raising two questions: how hospitals allocate resources to end devices and how other agencies participating in this program will compensate for the resources used by other hospitals.

    “The authors in [69] find the solution to this problem by arising two questions: how hospitals allocate resources to end devices and how other agencies participating in this program will compensate for the resources used by other hospitals.”
    Quote in the paperNot a substantial claim0 of 1 sources checked
    • cites [69] source

      Not checked in the free run.

    • Rests on an assumption: In vanilla federated learning algorithms, all IoMT devices must participate in training and share their aggregation model, but in most cases this does not occur.
    • Rests on an assumption: Due to limited computation capability of IoMT devices and privacy and trustworthiness of third-party platforms, IoMT devices do not show willingness to contribute to federated learning approaches.

  • Probably not a substantial claim: may not pass S3 (load-bearing).

    Interpretationreasoning

    The communication network is a crucial factor when deploying FL because model updates are often transmitted through it.

    “The communication network is a crucial factor to consider when deploying FL for applications since model updates are often transmitted through it.”
    Quote in the paperNot a substantial claim

  • Probably not a substantial claim: may not pass S3 (load-bearing).

    Premisecitation

    In the dynamic FL algorithm, the central server decides which client's model will be selected for aggregation based on updating time.

    “Furthermore, the central server decides which client's model will be selected for aggregation based on updating time.”
    Quote in the paperNot a substantial claim0 of 1 sources checked
    • cites [79] source

      Not checked in the free run.

  • Probably not a substantial claim: may not pass S3 (load-bearing).

    Premisecitation

    In the dynamic FL algorithm, during each iteration, the hospital decides whether to participate in training based on model performance.

    “During each iteration, the hospital decides whether to participate in training or not based on the model performance.”
    Quote in the paperNot a substantial claim0 of 1 sources checked
    • cites [79] source

      Not checked in the free run.

  • Probably not a substantial claim: may not pass S3 (load-bearing).

    Premisecitation

    A deep reinforcement learning based algorithm is presented where DRL is responsible for allocation awards at the aggregator and selecting devices to suggest the learning model for their data.

    “To tackle that drawback, a deep reinforcement learning (DRL) based algorithm is presented in [68], where DRL is responsible for allocation awards at the aggregator and selecting devices to suggest the learning model for their data.”
    Quote in the paperNot a substantial claim0 of 1 sources checked
    • cites [68] source

      Not checked in the free run.

    • Rests on an assumption: In vanilla federated learning algorithms, all IoMT devices must participate in training and share their aggregation model, but in most cases this does not occur.
    • Rests on an assumption: Due to limited computation capability of IoMT devices and privacy and trustworthiness of third-party platforms, IoMT devices do not show willingness to contribute to federated learning approaches.

  • Probably not a substantial claim: may not pass S3 (load-bearing). May combine two claims.

    Premisecitation

    The problem is treated as non-convex optimization, and distributed techniques are investigated to solve such problems.

    “In this context, the problem is treated as non-convex optimization, and distributed techniques are investigated to solve such problems.”
    Quote in the paperNot a substantial claim0 of 1 sources checked
    • cites [69] source

      Not checked in the free run.

    • Rests on an assumption: In vanilla federated learning algorithms, all IoMT devices must participate in training and share their aggregation model, but in most cases this does not occur.
    • Rests on an assumption: Due to limited computation capability of IoMT devices and privacy and trustworthiness of third-party platforms, IoMT devices do not show willingness to contribute to federated learning approaches.

  • Probably not a substantial claim: may not pass S3 (load-bearing).

    Premisecitation

    In the dynamic FL algorithm, if the global server does not receive local model parameters at the appropriate time, that model is excluded from global aggregation.

    “In case if a global server does not receive the local model parameters at the appropriate time, then that model is excluded from global aggregation.”
    Quote in the paperNot a substantial claim0 of 1 sources checked
    • cites [79] source

      Not checked in the free run.

  • May combine two claims.

    Premisecitation

    Collection and sharing of abundant data for designing an effective training model is challenging due to privacy concerns and most users may not share their medical records for analysis.

    “However, the collection and sharing of abundant data for designing an effective training model is a challenging task due to privacy concerns and the fact that most user may not share their medical records for analysis purposes [77].”
    Quote in the paperSubstantial claim0 of 1 sources checked
    • cites [77] Choudhury, 2019, Differential privacy-enabled federated learning for sensitive health data

      Not checked in the free run.

  • Probably not a substantial claim: may not pass S3 (load-bearing). May combine two claims.

    Premisecitation

    LSTM-based anomaly detection was built-in [71] and FL-based digital twin was used for preventing adversaries from getting private information.

    “The long short term memory (LSTM) based anomaly detection was built-in [71] and FL based DT was used for preventing adversaries from getting private information.”
    Quote in the paperNot a substantial claim0 of 1 sources checked
    • cites [71] source

      Not checked in the free run.

  • Probably not a substantial claim: may not pass S3 (load-bearing), S4 (based).

    Interpretationreasoning

    The communication issue is more severe in smart healthcare systems where a large number of IoMT devices are present.

    “The case scenario is more severe in the case of a smart healthcare system where a large number of IoMT devices are present.”
    Quote in the paperNot a substantial claim

  • May combine two claims.

    Interpretationreasoning

    FL-based algorithms are the most viable option for medical image processing because the model is trained on multiple datasets without sharing private information.

    “In such a case scenario, FL-based algorithms are considered to be the most viable option, as the model is trained on multiple datasets without sharing their private information.”
    Quote in the paperSubstantial claim
    • Rests on an assumption: Many institutions avoid sharing their private data for processing due to privacy concerns.

  • Probably not a substantial claim: may not pass S3 (load-bearing). May combine two claims.

    Interpretationcitation

    Through digital twins, a virtual environment containing patients can be created and doctors can test prescribed medication in a virtual environment before using it on real-world patients.

    “From the IoMT perspective, through DT virtual environment containing patients can be created and the doctors can test their prescribed medication in a virtual environment before using them on real-world patients.”
    Quote in the paperNot a substantial claim

  • May combine two claims.

    Interpretationreasoning

    FL-based techniques provide a more reliable intelligent solution to EHR management while preserving the privacy of patient data, especially where multiple cooperation data is involved.

    “In such circumstances, FL-based techniques provide a more reliable intelligent solution to EHR management while preserving the privacy of patient data, especially where multiple cooperation data is involved.”
    Quote in the paperSubstantial claim
    • Rests on an assumption: Traditional ML techniques have privacy concerns that could lead to sensitive data leakage during analysis.

  • Probably not a substantial claim: may not pass S3 (load-bearing).

    Premisecitation

    Many professionals are working on the deployment and development of a 6G wireless network.

    “Even though the full potential of a 5G network is yet to be recognized by the world, many professionals are working on the deployment and development of a 6G wireless network [82].”
    Quote in the paperNot a substantial claim0 of 1 sources checked
    • cites [82] source

      Not checked in the free run.

  • May combine two claims.

    Premisereasoning

    Many blockchain algorithms are presented for IoMT to reduce the functionality of the central server, but no comparative analysis has been found.

    “For instance, to reduce the functionality of the central server, many blockchain algorithms are presented for IoMT, but no comparative analysis has been found.”
    Quote in the paperNot a substantial claim0 of 1 sources checked
    • cites [80] source

      Not checked in the free run.

  • Probably not a substantial claim: may not pass S4 (based).

    Interpretationnone given

    Deployment of smart healthcare systems in the real world is limited due to the lack of privacy protection.

    “Despite the many potential applications of smart healthcare systems, their deployment in the real world is limited due to the lack of privacy protection.”
    Quote in the paperNot a substantial claim

  • Classified as probably not a substantial claim. May combine two claims.

    Interpretationreasoning

    FL is a viable option for privacy in smart healthcare because FL operates collaboratively and provides privacy to user information.

    “For this specific purpose, FL is considered to be a viable option as FL operates in a collaborative fashion and provides privacy to user information.”
    Quote in the paperSubstantial claim

  • Probably not a substantial claim: may not pass S3 (load-bearing).

    Premisecitation

    The adoption of the 6G network can be used in many industrial application sectors ranging from Industry 5.0 to body area networks.

    “The adoption of the 6G network can be used in many industrial application sectors ranging from Industry 5.0 to body area networks.”
    Quote in the paperNot a substantial claim0 of 1 sources checked
    • cites [82] source

      Not checked in the free run.

  • Probably not a substantial claim: may not pass S3 (load-bearing).

    Premisecitation

    In a 6G network, data generation and collection will be abundant.

    “In a 6G network, the data generation and collection will be abundant.”
    Quote in the paperNot a substantial claim0 of 1 sources checked
    • cites [82] source

      Not checked in the free run.

42 of 42

  • The paper relies on this without support.

    Electronic healthcare systems are considered an easy target for attackers due to poor defense systems for information security.

    “Due to poor defense systems for information security, electronic healthcare systems are considered an easy target for attackers.”

  • The paper relies on this without support.

    Constraints associated with individual IoT devices, such as power and resources, limit federated learning operation.

    “Moreover, there are constraints associated with individual IoT devices, such as power and resources, limiting the FL operation.”

  • The paper relies on this without support.

    Insufficient data makes it impossible to train an accurate model because input data is imbalanced and lacks enough feature information.

    “Moreover, insufficient data makes it impossible to train an accurate model since the input data is imbalanced and lacks enough feature information.”

  • The paper relies on this without support.

    Training of networks on diverse data increases the generalization capability of federated learning.

    “Moreover, the training of networks on diverse data increases the generalization capability v of FL.”

  • The paper relies on this without support.

    If an adversary gets access to the central server, user information can be extracted.

    “For example, if an adversary gets access to the central server, then user information can be extracted, as shown in Fig. 1.”

  • The paper relies on this without support.

    In distributed federated learning, model updates communicated with central servers by IoMT devices keep leaking some training data information.

    “In distributed FL, the model updates communicated with central servers by IoMT devices keep leaking some training data information.”

  • The paper relies on this without support.

    By initiating a differential attack, the adversary can find out whether end devices are dedicated to a particular task or not.

    “By initiating the differential attack, the adversary can find out whether the end devices are dedicated to a particular task or not.”

  • The paper relies on this without support.

    The major known technique mostly used to enhance privacy preservation in many areas is based on differential privacy.

    “The major known technique that is mostly used to enhance privacy preservation in many areas is based on differential privacy.”

  • The paper relies on this without support.

    In vanilla federated learning algorithms, all IoMT devices must participate in training and share their aggregation model, but in most cases this does not occur.

    “In vanilla FL algorithms, all the IoMT devices must participate in the training process and share their aggregation model, but in most cases, this does not occur.”

  • The paper relies on this without support.

    Due to limited computation capability of IoMT devices and privacy and trustworthiness of third-party platforms, IoMT devices do not show willingness to contribute to federated learning approaches.

    “Due to the limited computation capability of IoMT devices and the privacy and trustworthiness of third-party platforms, IoMT devices do not show any willingness to contribute to FL approaches.”

  • The paper relies on this without support.

    Traditional ML techniques have privacy concerns that could lead to sensitive data leakage during analysis.

    “However, traditional ML techniques have privacy concerns, which could lead to sensitive data being leaked during analysis.”

  • The paper relies on this without support.

    Many institutions avoid sharing their private data for processing due to privacy concerns.

    “However, due to privacy concerns, many institutions avoid sharing their private data for processing.”

  • The paper relies on this without support.

    In real-world applications, the data available by each client is unique and heterogeneous.

    “In real-world applications, the data available by each client is unique; that is, it might be in different forms like images, video, or text, and they will be showing different information like blood type and sugar level, etc.”

  • The paper relies on this without support.

    A limited comprehensive survey regarding FL application in IoMT is available.

    “However, given the fact of the importance of FL, a limited comprehensive survey regarding FL application in IoMT is available.”
14 of 14

Claims other papers make about this one

Each is checked against this paper’s own text: does it say what the citing paper says it does?

  • The target of federated learning in health care AI applications is to preserve the privacy of sensitive patient information communicated between hospitals and end users, particularly through Internet of Medical Things devices.

    “The target of FL in health care AI applications is to preserve the privacy of sensitive patient information communicated between hospitals and end users, particularly through Internet of Medical Things (IoMT) devices.”

    From Generative AI in medical practice: in-depth exploration of privacy and security challenges (2024) · cites this paper as [49]

    The Introduction says FL protects privacy in smart-healthcare networks by training across IoMT networks without accessing local data, mitigating disclosure of sensitive patient information. The copy does not specifically describe FL’s target as protecting information communicated between hospitals and end users; that communication path is more specific than the cited passage supports. (LLM, conf=0.83)

    Experimental — accuracy not yet measured

  • Federated learning provides privacy to end users by only sharing gradients during training.

    “It provides privacy to end users by only sharing gradients during training.”

    From Generative AI in medical practice: in-depth exploration of privacy and security challenges (2024) · cites this paper as [49]

    The Introduction (page unavailable) says federated learning averages local model updates without accessing local data, mitigating risks to sensitive information. It does not say that only gradients are shared, and the “Privacy data leakage attack” section warns that communicated model updates can leak training-data information. Thus, the source supports the privacy benefit but not the claim that sharing only gradients provides privacy. (LLM, conf=0.88)

    Experimental — accuracy not yet measured

  • Implementing federated learning in Internet of Medical Things networks presents challenges, such as the need for robust federated learning for diffused health data sets, integration of federated learning with next-generation Internet of Medical Things networks, and use of blockchain for decentralized and secure data storage.

    “However, implementing FL in IoMT networks presents challenges, such as the need for robust FL for diffused health data sets, the integration of FL with next-generation IoMT networks, and the use of blockchain for decentralized and secure data storage.”

    From Generative AI in medical practice: in-depth exploration of privacy and security challenges (2024) · cites this paper as [49]

    Split into 2 atoms — 0 confirmed, 2 partially supported, 0 disputed, 0 misattributed, 0 unverifiable. Expand the claim to see per-atom verdicts.

    Experimental — accuracy not yet measured

  • Federated learning is a distributed AI paradigm that offers privacy preservation in smart health care systems by allowing models to be trained without accessing the local data of participants.

    “FL is a distributed AI paradigm that offers privacy preservation in smart health care systems by allowing models to be trained without accessing the local data of participants.”

    From Generative AI in medical practice: in-depth exploration of privacy and security challenges (2024) · cites this paper as [49]

  • The federated learning approach incorporates advanced techniques such as reinforcement learning, digital twin, and GANs to detect and prevent privacy threats in Internet of Medical Things networks.

    “The approach incorporates advanced techniques such as reinforcement learning, digital twin, and GANs to detect and prevent privacy threats in IoMT networks.”

    From Generative AI in medical practice: in-depth exploration of privacy and security challenges (2024) · cites this paper as [49]

Self-check

Percentages, sums and reported differences re-computed from the paper’s own numbers.

Each contradiction between two passages listed here was confirmed by an independent check.

  • Section II is assigned two different topics

    Minor

    The organization paragraph says both the security and privacy discussion and the FL architecture discussion are in Section II. The FL discussion is headed Section III.

  • X-ray example cites a CT-imaging study

    Minor

    The text cites [78] for training on chest X-ray images, but the cited reference is titled as a study of COVID-19 detection using CT imaging.

2 of 2

The citation lists haven’t been fetched yet

References 109 · Cited by 366

Research works printed but not linked: 4

  1. Meulen, 2017, Gartner says 8.4 billion connected Things will be in use in 2017, up 31 percent from 2016, Gartner Letzte Aktual, 7
  2. McMahan, 2017, Communication-efficient learning of deep networks from decentralized data, 1273
  3. 10.1109/JSEN.2021.3076767 · DOI · Paper page
  4. Gadekallu, 2022, Blockchain for the metaverse: A review

Authors
Ali, M., Naeem, F., Tariq, M., & Kaddoum, G.
Affiliations
  • École de Technologie Supérieure (CA)
  • National University of Computer and Emerging Sciences (PK)
ORCID
Venue
IEEE Journal of Biomedical and Health Informatics
Publisher
Institute of Electrical and Electronics Engineers (IEEE)
Published
2023-02
Volume
27
Issue
2
Pages
778-789
ISSN
  • 2168-2194 (ISSN-L)
  • 2168-2208
DOI
10.1109/jbhi.2022.3181823
OpenAlex
W4221148884
Semantic Scholar
3c6f7ec37cedd2d487c42ce58a6da2c9293cbdd9
Cited by
366
References
109
Citations by year
2026: 109 · 2025: 125 · 2024: 79 · 2023: 48 · 2022: 5
FWCI
21.06
Type
journal article
Language
en
Open access
Licence
Locations
  • IEEE Journal of Biomedical and Health Informatics (publishedVersion)
  • PubMed (publishedVersion)
  • arXiv (Cornell University) (submittedVersion)
  • Espace ÉTS (ETS) (submittedVersion)
  • arXiv (Cornell University)
Topics
  • Physical Sciences › Computer Science › Artificial Intelligence › Privacy-Preserving Technologies in Data
  • Physical Sciences › Computer Science › Information Systems › Internet of Things and AI
  • Physical Sciences › Computer Science › Computer Networks and Communications › IoT and Edge/Fog Computing
Keywords
federated learning, privacy preservation, smart healthcare, Internet of Medical Things, wearable devices, privacy risks, deep reinforcement learning, digital twin, generative adversarial networks, distributed learning
MeSH terms
Artificial Intelligence, Communication, Electronics, Hospitals, Humans, Privacy

About these checks

Registry, retraction, Cites and statistics results are computed, not judged, with one exception: a printed reference that no index lists is read once by a model before it counts as a citation error. Claims are extracted by a model and kept only when their quote appears verbatim in the text; each cited source is then checked, with a model’s help, for whether it says what the claim relies on. Those verdicts can be wrong — read the claim’s quote and the cited source before relying on one.

Claims read by deepseek-v4.1-flash under definition version 3 (extract-v3) on 2026-10-09.

Classified by jev-1.13-20260917 (gold-guideline-v0.2) on 2026-10-09.

Statistics and arithmetic are recomputed, not judged.

Confidence bands are shown only for fields calibrated on our annotated test set. None is calibrated yet, so no band is shown.

What counts as a claim ›How we measure confidence ›How we check this ›

Are you an author, or do you see a problem with this page? Request a review