CoChat
Sign Up
Cited by 24Artificial Intelligence in HealthcareMachine Learning in HealthcareDiabetes Management and Research

Artificial intelligence based prediction models for individuals at risk of multiple diabetic complications: A systematic review of the literature

Open access
Gosak, Lucija, Martinović, Kristina, Lorber, Mateja, Stiglic, Gregor·2022·Journal of Nursing Management
DOI 10.1111/jonm.13894PMC PMC10100477
Open this paper in CoChat — free accountOpen PDF

Abstract

AIM: The aim of this review is to examine the effectiveness of artificial intelligence in predicting multimorbid diabetes-related complications. BACKGROUND: In diabetic patients, several complications are often present, which have a significant impact on the quality of life; therefore, it is crucial to predict the level of risk for diabetes and its complications. EVALUATION: International databases PubMed, CINAHL, MEDLINE and Scopus were searched using the terms artificial intelligence, diabetes mellitus and prediction of complications to identify studies on the effectiveness of artificial intelligence for predicting multimorbid diabetes-related complications. The results were organized by outcomes to allow more efficient comparison. KEY ISSUES: Based on the inclusion/exclusion criteria, 11 articles were included in the final analysis. The most frequently predicted complications were diabetic neuropathy (n = 7). Authors included from two to a maximum of 14 complications. The most commonly used prediction models were penalized regression, random forest and Naïve Bayes model neural network. CONCLUSION: The use of artificial intelligence can predict the risks of diabetes complications with greater precision based on available multidimensional datasets and provides an important tool for nurses working in preventive health care. IMPLICATIONS FOR NURSING MANAGEMENT: Using artificial intelligence contributes to a better quality of care, better autonomy of patients in diabetes management and reduction of complications, costs of medical care and mortality.

AI snapshot

Written from the paper’s full text.

TL;DR

This systematic review examined AI-based prediction models for multiple diabetes-related complications and included 11 studies. It found that diabetic neuropathy and nephropathy were the most frequently predicted complications, with best-model AUCs generally ranging from 0.685 to 0.847 when excluding one near-perfect result.

Key findings

  • Of 251 records, 38 addressed prediction of diabetes-related complications, but only 11 studies predicted two or more complications; models included from 2 to a maximum of 14 complications.
  • The most frequently predicted complications were diabetic neuropathy (n = 7), diabetic nephropathy (n = 6), and diabetic retinopathy (n = 5). The most common AI approaches were penalized regression, random forest, Naïve Bayes, and neural network models.
  • Reported performance varied: Dagliati et al. reported 3-year AUCs of 0.489 for SVM, 0.5 for RF, 0.726 for LR, and 0.533 for NB; Fan et al. reported AUCs of 0.847 ± 0.081 for XF, 0.787 ± 0.081 for CHAID, and 0.720 ± 0.060 for QUEST; Liu et al. reported AUCs of 0.545 for BN, 0.685 for BN-wopi, 0.505 for NB, 0.516 for RF, and 0.527 for C5.0.
  • Excluding Ozdemir et al.’s near-perfect 0.99 result, the best model AUC in each study ranged from 0.685 to 0.847. Ljubic et al. reported four-visit accuracies of 0.746 ± 0.053 for GRU RNN, 0.746 ± 0.053 for bidirectional GRU, 0.719 ± 0.073 for 1-way LSTM, 0.671 ± 0.033 for RF, and 0.668 ± 0.039 for MLP.

Methods

  • Systematic review conducted according to Khan et al. (2003); PubMed, CINAHL, MEDLINE, and Scopus were searched using terms for artificial intelligence, diabetes mellitus type 2, and prediction of complications, with no publication date limits.
  • Inclusion criteria covered quantitative, qualitative, and mixed-method studies in English involving patients already diagnosed with diabetes and AI techniques such as SVM, decision tree, random forest, gradient boosting, and neural networks for predicting multimorbid complications; studies using exclusively regression models were excluded, though regression results were extracted when used as baseline comparisons.
  • Two authors assessed eligibility and extracted data, resolving disagreements by discussion. The review extracted AUC with confidence intervals or standard deviations where available, and accuracy, sensitivity, and specificity where AUC was not reported.

Limitations

  • The paper states its biggest limitation is heterogeneity: included studies used different methodological approaches and reported results that could not be synthesized due to inequalities, and population sizes ranged widely from 129 to 287,438, making comparison difficult.
  • The paper states that different cross-validation and internal versus external evaluation approaches can influence reliability, and that in many cases the evaluation protocol description was weak or nonexistent; Ozdemir et al. provided no data or code, making robustness difficult to assess.
  • Evident from the text: only 11 studies were included and no quantitative meta-analysis was performed; the English-language restriction may also limit generalizability.

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

Integrity signals

Registry

Crossref · 10.1111/jonm.13894

Cites

51 of 51 references verified0 citation errors · 0 not checkable

Statistics

Nothing to recompute.

Claims the paper makes

23 of 33 cited sources checked · 11 supported · 0 flagged · accuracy not yet measured

  • Findingcitation

    In Dagliati et al. (2018), logistic regression after random forest imputation identified risk factors for microvascular complications.

    “LR after RF imputation identified risk factors for microvascular complications.”
    Quote in the paperNot a substantial claim1 of 1 sources checked
    • cites Dagliati et al. (2018) source
  • Premisecitation

    Predicting the occurrence of different complications is a challenging task due to different risk factors, unbalanced data, and rapid changes.

    “but predicting the occurrence of different complications is a challenging task due to different risk factors, unbalanced data and rapid changes (Singla et al., 2019).”
    Quote in the paperSubstantial claim1 of 1 sources checked
    • cites Singla et al., 2019 source

      The Abstract (page not specified) says AI/ML is used to predict the risk of complications such as nephropathy and retinopathy. The Type 2 Diabetes Management section discusses noisy data and individual factors in treatment decisions, but neither passage says complication prediction is challenging because of different risk factors, unbalanced data, or rapid changes. (LLM, conf=0.76)

      Experimental — accuracy not yet measured
  • Premisecitation

    AI plays an important role in predicting complications using basic clinical and biochemical patient data.

    “AI plays an important role in predicting complications using basic clinical and biochemical patient data”
    Quote in the paperSubstantial claim1 of 1 sources checked
    • cites Singla et al., 2019 source

      The section labeled Abstract (page unavailable) says AI/ML is used to predict the risk of diabetes complications, including nephropathy and retinopathy. It does not specify that those predictions use basic clinical and biochemical patient data; the blood-glucose and HbA1c measures discussed under Type 2 Diabetes Management concern treatment decisions, not complication prediction. (LLM, conf=0.86)

      Experimental — accuracy not yet measured
  • Findingcitation

    Liu et al. (2020) presented AUC for BN (0.545), BN-wopi (0.685), NB (0.505), RF (0.516), and C5.0 (0.527).

    “Liu et al. (2020) presented the AUC for the following models: BN (0.545), BN‐wopi (0.685), NB (0.505), RF (0.516) and C5.0 (0.527).”
    Quote in the paperNot a substantial claim1 of 1 sources checked
    • cites Liu et al. (2020) source

      Section 3.1 compares BN, BN-wopi, NB, RF and C5.0 using AUC, and the paper reports an AUC of 0.545 for predicting DPN. The passages available from our copy do not verify the claimed model-specific values of 0.685, 0.505, 0.516 and 0.527, or explicitly associate 0.545 with BN. (LLM, conf=0.78)

      Experimental — accuracy not yet measured
  • Findingcitation

    In Ljubic et al. (2020), the RNN GRU model was more accurate (73%-83% vs. 66%-76%) compared with traditional models.

    “The RNN GRU model was more accurate (73%–83% vs. 66%–76%) compared with traditional models.”
    Quote in the paperNot a substantial claim1 of 1 sources checked
    • cites Ljubic et al. (2020) source
  • Classified as probably not a substantial claim. 22
  • May combine two claims. 19
  • The quote only partly states this claim. 2

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

    Findingcitation

    Lee et al. (2021) found that higher HbA1c and lipid measurements were associated with an increased risk of complications and various comorbidities.

    “Lee et al. (2021) found that higher HbA1c and lipid measurements were associated with an increased risk of complications and various comorbidities.”
    Quote in the paperNot a substantial claim1 of 1 sources checked
    • cites Lee et al. (2021) source

      The Abstract’s Results and the Conclusion (page numbers unavailable) associate higher HbA1c and lipid *variability* with complications and other adverse outcomes, while the claim says higher measurements. The Results section reports that elevated mean total cholesterol predicts most adverse outcomes, but higher HDL-C generally lowers risk; the Discussion also describes non-linear relationships for HbA1c and lipid levels. The claim therefore overgeneralizes the findings about measurement levels. (LLM, conf=0.88)

      Experimental — accuracy not yet measured

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

    Premisecitation

    AI supports nurses in clinical decision-making and other tasks that are not directly related to the patient.

    “AI supports nurses in clinical decision‐making and other tasks that are not directly related to the patient (Seibert et al., 2021).”
    Quote in the paperNot a substantial claim1 of 1 sources checked
    • cites Seibert et al., 2021 source

  • May combine two claims.

    Findingcitation

    Ljubic et al. (2020) did not present AUC but gave accuracy for all models at four, three, and two visits.

    “Ljubic et al. (2020) did not present the AUC in their paper but only gave the accuracy for all models at four, three and two visits.”
    Quote in the paperNot a substantial claim1 of 1 sources checked
    • cites Ljubic et al. (2020) source

      Materials and Methods identifies accuracy as the evaluation metric, and the Results report accuracy for all four models at two, three, and four visits. The claim that the paper gave only accuracy is too broad: Results Table 3 also reports sensitivity and specificity for the bidirectional GRU model in the four-visit scenario. The searched passages did not show an AUC result. (LLM, conf=0.88)

      Experimental — accuracy not yet measured

  • May combine two claims.

    Findingcitation

    Fan et al. (2021) reported AUC of 0.847 ± 0.081 for XF, 0.787 ± 0.081 for CHAID, and 0.720 ± 0.060 for QUEST.

    “Fan et al. (2021) reported AUC of 0.847 ± 0.081 for XF, 0.787 ± 0.081 for CHAID and 0.720 ± 0.060 for QUEST.”
    Quote in the paperNot a substantial claim1 of 1 sources checked
    • cites Fan et al. (2021) source

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

    Findingcitation

    Ljubic et al. (2020) concluded that the RNN GRU model was the best of the analysed RNN GRU models, followed by the RNN LSTM.

    “Based on the results, the authors concluded that it is the best of the analysed RNN GRU models, followed by the RNN LSTM.”
    • Also stated in: § RESULTS · plain
      “Ljubic et al. (2020) cite the RNN GRU model as the most accurate model, followed by the RNN LSTM model.”
    Quote in the paperNot a substantial claim1 of 1 sources checked
    • cites Ljubic et al. (2020) source

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

    Premisecitation

    Nurses, as the largest part of all workers involved in health systems, will benefit enormously from AI.

    “Nurses, as the largest part of all workers involved in health systems, will benefit enormously from AI (Shang, 2021).”
    Quote in the paperNot a substantial claim1 of 1 sources checked
    • cites Shang, 2021 source

      The Introduction and background says nurses are the largest proportion of healthcare workers and would theoretically benefit the most from AI. The citing sentence presents that projected benefit as certain, whereas the source qualifies it and notes that adoption of AI in nursing remains slow. (LLM, conf=0.86)

      Experimental — accuracy not yet measured

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

    Findingcitation

    Aminian et al. (2020) found that cardiovascular events are among the most frequent complications in patients with type 2 diabetes and obesity.

    “They found that cardiovascular events are among the most frequent complications in patients with type 2 diabetes and obesity.”
    Quote in the paperNot a substantial claim1 of 1 sources checked
    • cites Aminian et al. (2020) source

      The Introduction (page unavailable) identifies cardiovascular outcomes as important complications in patients with type 2 diabetes and obesity, and its baseline-characteristics table reports coronary artery disease and heart failure in roughly 10% of the study groups. The paper does not establish that cardiovascular events, as a group, are among the most frequent complications; that frequency ranking goes beyond what these passages show. (LLM, conf=0.80)

      Experimental — accuracy not yet measured

  • May combine two claims.

    Findingcitation

    For four visits in Ljubic et al. (2020), the accuracy was 0.746 ± 0.053 for GRU RNN, 0.746 ± 0.053 for bidirectional GRU, 0.719 ± 0.073 for 1-way LSTM, 0.671 ± 0.033 for RF, and 0.668 ± 0.039 for MLP.

    “We present a comparison for four visits only and the values for the GRU RNN model with the accuracy of 0.746 ± 0.053, bidirectional GRU with 0.746 ± 0.053, 1‐way LSTM with 0.719 ± 0.073, RF with 0.671 ± 0.033 and MLP with 0.668 ± 0.039.”
    Quote in the paperNot a substantial claim1 of 1 sources checked
    • cites Ljubic et al. (2020) source

      The Results (Table 2) report exactly these four-visit accuracies for neuropathy: 0.746 ± 0.053 for bidirectional GRU, 0.719 ± 0.073 for 1-way LSTM, 0.671 ± 0.033 for RF, and 0.668 ± 0.039 for MLP. The paper calls the bidirectional GRU a GRU RNN; it does not report a separate GRU RNN model with the same accuracy. The claim presents one model as two. (LLM, conf=0.88)

      Experimental — accuracy not yet measured

  • May combine two claims.

    Findingcitation

    In Fan et al. (2021), XF was the best performing model for predicting diabetic nephropathy and diabetic angiopathy, D for diabetic neuropathy and diabetic eye disease, and BN for HbA1c.

    “XF was the best performing model for predicting diabetic nephropathy and diabetic angiopathy, D for predicting diabetic neuropathy and diabetic eye disease and BN for HbA1c (Fan et al., 2021).”
    Quote in the paperNot a substantial claim1 of 1 sources checked
    • cites Fan et al. (2021) source

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

    Findingcitation

    In Ljubic et al. (2020), RNN models were most accurate in predicting depressive disorder and IHD.

    “RNN models were most accurate in predicting depressive disorder and IHD.”
    Quote in the paperNot a substantial claim1 of 1 sources checked
    • cites Ljubic et al. (2020) source

      The Discussion says RNN models were most accurate when predicting depressive disorder and ICHD. The cited paper specifies chronic ischemic heart disease, whereas the claim says IHD without that qualifier. (LLM, conf=0.86)

      Experimental — accuracy not yet measured

  • Probably not a substantial claim: may not pass S2 (checkable), S3 (load-bearing).

    Premisecitation

    Accurate prediction helps to target nursing interventions better.

    “Consequently, accurate prediction helps to target nursing interventions better (Ljubic et al., 2020).”
    Quote in the paperNot a substantial claim1 of 1 sources checked
    • cites Ljubic et al., 2020 source

      The Discussion (page not available) says accurate prediction can identify high-risk patients for monitoring and intervention, supporting the general point about targeting interventions. It does not specifically discuss nursing interventions, and it frames the benefit as a potential application rather than an established effect. (LLM, conf=0.84)

      Experimental — accuracy not yet measured

  • May combine two claims.

    Findingcitation

    Shi et al. (2020) found a moderately good differentiation and ability to confirm and predict the incidence of diabetic nephropathy and diabetic retinopathy.

    “The authors found a moderately good differentiation and ability to confirm and predict the incidence of diabetic nephropathy and diabetic retinopathy.”
    Quote in the paperNot a substantial claim1 of 1 sources checked
    • cites Shi et al. (2020) source

  • May combine two claims.

    Findingcitation

    For the 3-year forecasting models in Dagliati et al. (2018), the AUC was 0.489 for SVM, 0.5 for RF, 0.726 for LR, and 0.533 for NB.

    “We compared the AUCs for the 3‐year forecasting models, and the AUCs were 0.489 for the SVM model, 0.5 for the RF model, 0.726 for the LR and 0.533 for the NB model.”
    Quote in the paperNot a substantial claim1 of 1 sources checked
    • cites Dagliati et al. (2018) source

      The Results section says the paper evaluated AUC by forecasting scenario and modeling strategy, including three-year models. However, the available passages do not substantiate any of the four quoted three-year AUC values; Table 4 lists different AUCs for the final three-year models. Searches for the quoted values and Table 3 did not retrieve the numerical comparison. (LLM, conf=0.78)

      Experimental — accuracy not yet measured

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

    Findingown data

    In Dagliati et al. (2018), SD or CIs were not reported.

    “SD or CIs were not reported.”
    Quote in the paperNot a substantial claim1 of 1 sources checked
    • cites Dagliati et al. (2018) source

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

    Findingcitation

    In Dagliati et al. (2018), accuracy was reported only for LR.

    “Accuracy was reported only for LR.”
    Quote in the paperNot a substantial claim1 of 1 sources checked
    • cites Dagliati et al. (2018) source

  • The quote only partly states this claim.

    Findingown data

    The included studies predicted from two to a maximum of 14 diabetes-related complications.

    “Authors included from two to a maximum of 14 complications.”
    Quote in the paperNot a substantial claim

  • May combine two claims.

    Findingcitation

    Lagani et al. (2015) used 51 clinical variables and included between five and fifteen risk factors depending on the specific outcome.

    “Fifty‐one clinical variables were used in the selected predictive models and included between five and fifteen risk factors depending on the specific outcome (Lagani et al., 2015).”
    Quote in the paperNot a substantial claim0 of 1 sources checked
    • cites Lagani et al. (2015) source

  • May combine two claims.

    Findingcitation

    Lagani et al. (2015) reported AUC for predicting neuropathy in both type 1 and type 2 diabetes: T1DM (internal) 0.6661, T1DM (external) 0.735, and T2DM (external) 0.4359.

    “Lagani et al. (2015) reported AUC for predicting neuropathy in both type 1 and type 2 diabetes. T1DM (internal) was 0.6661, T1DM (external) was 0.735 and T2DM (external) was 0.4359.”
    Quote in the paperNot a substantial claim0 of 1 sources checked
    • cites Lagani et al. (2015) source

  • May combine two claims.

    Findingcitation

    Ozdemir et al. (2020) reported the same result for AUC and accuracy, resulting in an almost perfect classification performance of 0.99.

    “Ozdemir et al. (2020) reported the same result for AUC and accuracy resulting in an almost perfect classification performance of 0.99.”
    Quote in the paperNot a substantial claim0 of 1 sources checked
    • cites Ozdemir et al. (2020) source

  • May combine two claims.

    Findingcitation

    In Wang et al. (2021), the accuracy for covering all three complications at once was 0.685 ± 0.000 for BR (linear), 0.701 ± 0.000 for BR (poly), 0.701 ± 0.000 for BR (rbf), 0.520 ± 0.001 for WML-SSLM (linear), 0.661 ± 0.002 for WML-SSLM (poly), and 0.697 ± 0.002 for WML-SSLM (rbf).

    “BR (linear) resulted in accuracy of 0.685 ± 0.000, BR (poly) of 0.701 ± 0.000, BR (rbf) of 0.701 ± 0.000, WML‐SSLM (linear) of 0.520 ± 0.001, WML‐SSLM (poly) of 0.661 ± 0.002 and WML‐SSLM (rbf) of 0.697 ± 0.002.”
    Quote in the paperSubstantial claim0 of 1 sources checked
    • cites Wang et al. (2021) source

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

    Findingown data

    The search string found 251 results in the databases.

    “Based on the search string, we found a total of 251 results in the databases.”
    Quote in the paperNot a substantial claim

  • May combine two claims.

    Findingreasoning

    The included research reported different methodological approaches and different results that cannot be synthesized due to inequalities.

    “The included research reported different methodological approaches and, above all, reported different results that cannot be synthesized due to inequalities.”
    Quote in the paperSubstantial claim

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

    Interpretationreasoning

    AI-based prediction of diabetes complications provides an important tool for nurses working in preventive health care.

    “provides an important tool for nurses working in preventive health care.”
    Quote in the paperNot a substantial claim

  • Probably not a substantial claim: may not pass S4 (based). May combine two claims. The quote only partly states this claim.

    Interpretationreasoning

    Using artificial intelligence contributes to reduction of complications, costs of medical care and mortality.

    “Using artificial intelligence contributes to a better quality of care, better autonomy of patients in diabetes management and reduction of complications, costs of medical care and mortality.”
    Quote in the paperNot a substantial claim

  • Classified as probably not a substantial claim.

    Interpretationreasoning

    Using artificial intelligence contributes to a better quality of care.

    “Using artificial intelligence contributes to a better quality of care, better autonomy of patients in diabetes management and reduction of complications, costs of medical care and mortality.”
    Quote in the paperSubstantial claim

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

    Study design and conductcitation

    The literature review was conducted according to the recommendations of Khan et al. (2003).

    “The literature review was conducted according to the recommendations of Khan et al. (2003).”
    Quote in the paperNot a substantial claim0 of 1 sources checked
    • cites Khan et al. (2003)

      Couldn’t be matched to a paper in Crossref, so it can’t be checked.

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

    Study design and conductnone given

    The relevance and quality of the studies were assessed by two authors, who evaluated eligibility based on predefined inclusion and exclusion criteria.

    “The relevance and quality of the studies were assessed by two authors, who evaluated the eligibility of the articles based on predefined inclusion and exclusion criteria.”
    Quote in the paperSubstantial claim

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

    Study design and conductown data

    Two authors extracted the data from the included articles and displayed them using an identification table.

    “In the last two steps, two authors extract the data from the articles and display them using an identification table.”
    Quote in the paperNot a substantial claim

  • Probably not a substantial claim: may not pass S2 (checkable), S3 (load-bearing).

    Premisecitation

    With the introduction of technology, nurses' experience, knowledge and skills will be transformed into learning new ways of thinking and processing information.

    “With the introduction of technology, nurses' experience, knowledge and skills will be transformed into learning new ways of thinking and processing information (Robert, 2019).”
    Quote in the paperNot a substantial claim0 of 1 sources checked
    • cites Robert, 2019 source

  • Probably not a substantial claim: may not pass S2 (checkable), S4 (based). May combine two claims.

    Interpretationreasoning

    By using AI methods, we can help facilitate the control of diabetes and detect the presence of risk in patients for the development of multimorbid complications promptly.

    “By using AI methods, we can help facilitate the control of diabetes and detect the presence of risk in patients for the development of multimorbid complications promptly.”
    Quote in the paperNot a substantial claim

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

    Premisecitation

    The content of research in the field is very diverse.

    “The content of research in the field is very diverse”
    Quote in the paperNot a substantial claim0 of 1 sources checked
    • cites von Gerich et al., 2021

      Couldn’t be matched to a paper in Crossref, so it can’t be checked.

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

    Interpretationreasoning

    The use of AI methods serves as a tool for nurses when working with patients, making it easier to predict disease progression and contributing to better preventive care.

    “The use of AI methods also serves as a tool for nurses when working with patients, making it easier for them to predict disease progression and thus contributing better preventive care for patients.”
    Quote in the paperSubstantial claim

  • Probably not a substantial claim: may not pass S4 (based). May combine two claims.

    Interpretationreasoning

    The use of AI methods contributes to better quality of care, better autonomy of patients, and reduction of complications, costs of medical care, and mortality.

    “In this way, we contribute to a better quality of care, better autonomy of patients in the course of treatment of their disease and reduction of complications, costs of medical care and mortality.”
    Quote in the paperNot a substantial claim

  • Classified as probably not a substantial claim.

    Interpretationreasoning

    The use of AI provides a model for AI-assisted prognosis and diagnosis of next-generation diabetes complications.

    “and provides a model for AI‐assisted prognosis and diagnosis of next‐generation diabetes complications.”
    Quote in the paperSubstantial claim
34 of 34

  • The paper relies on this without support.

    The prediction performance metrics (AUC, accuracy, sensitivity, specificity) reported across the included studies are comparable enough to support a conclusion about the effectiveness of AI-based prediction models.

    “The results we aimed to extract from each study were areas under the curve (AUC) with corresponding confidence intervals (CIs) or standard deviations (SD). Since some studies did not report the AUC results, we also extracted accuracy, sensitivity and specificity where available.”
1 of 1

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.

  • Reported minimum sample size overlooks a 72-patient study

    Minor

    The discussion gives 129 as the minimum population size, but Table 2 includes Ozdemir et al. with a sample size of 72.

  • Included study uses exclusively regression models

    Minor

    The methods say studies using exclusively regression models were excluded, yet Table 2 includes Shi et al. and lists only LASSO regression and logistic regression for that study.

2 of 2

Cites

Flagged first, then by citations

Research works printed but not linked: 1

  1. Goyal R. &Jialal I.(2022).Diabetes mellitus type 2. StatPearls [Internet].https://www.ncbi.nlm.nih.gov/books/NBK513253/

Cited by

Sorted by citations

Related

Authors
Gosak, L., Martinović, K., Lorber, M., & Stiglic, G.
Affiliations
  • University of Maribor (SI)
  • University of Primorska (SI)
  • Faculty of Health Sciences University of Maribor Maribor Slovenia
Corresponding author
Gosak, Lucija
ORCID
Venue
Journal of Nursing Management
Publisher
Wiley
Published
2022-11
Volume
30
Issue
8
Pages
3765-3776
ISSN
  • 0966-0429 (ISSN-L)
  • 1365-2834
DOI
10.1111/jonm.13894
PMC
PMC10100477
OpenAlex
W4308179522
Cited by
24
References
51
Citations by year
2026: 5 · 2025: 7 · 2024: 7 · 2023: 4 · 2022: 1
FWCI
0.81
Type
journal article
Language
en
Open access
Licence
Locations
  • Journal of Nursing Management (publishedVersion, cc-by)
  • PubMed (publishedVersion)
  • University of Maribor digital library (University of Maribor) (publishedVersion, cc-by)
  • PubMed Central (submittedVersion, cc-by)
Topics
  • Health Sciences › Health Professions › Health Information Management › Artificial Intelligence in Healthcare
  • Physical Sciences › Computer Science › Artificial Intelligence › Machine Learning in Healthcare
  • Health Sciences › Medicine › Endocrinology, Diabetes and Metabolism › Diabetes Management and Research
Keywords
artificial intelligence, diabetes mellitus, diabetes complications, risk prediction model, diabetic neuropathy, multimorbidity, random forest, penalized regression, naive Bayes, artificial neural networks, nursing management, preventive medicine
MeSH terms
Artificial Intelligence, Bayes Theorem, Diabetes Mellitus, Humans, Quality of Life

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 2026-10-16.

Classified by jev-1.13-20260917 (gold-guideline-v0.2) on 2026-10-16.

Cited sources are checked by gpt-6-sol.

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