Envisioning Tools to Help Classify Type 1 Diabetes and Type 2 Diabetes in New-Onset Adult Diabetes

Abstract

A tool is needed to distinguish type 1 diabetes (T1D) and type 2 diabetes (T2D) in adults with new-onset diabetes because correct classification is needed for correct diagnoses and treatments. Current classification methods are usually applied to biomarkers using binary or quantitative classification with a cut point and may not be adequately nuanced. Combinations of clinical features are not necessarily specific for classifying and may not always indicate a single diagnosis. A probabilistic decision tree classification tool with multiple branches per decision node is needed for adults with new-onset diabetes to avoid misdiagnosis of actual T1D as T2D, misdiagnosis of actual T2D or monogenic diabetes as T1D, and misclassified patients in future population health studies which will lead to incorrect conclusions and suboptimal patient outcomes.

Keywords

diabetes type 1 diabetes type 2 diabetes new-onset adult diabetes decision tree machine learning

Introduction

Type 2 diabetes (T2D) is the most common type of diabetes in adults, making up about 90% to 95% of diabetes cases in adults.¹ However, there are also patients with adult-onset of type 1 diabetes (T1D). These patients may share phenotypic features with T2D patients, which leads to misclassification. It is estimated that 40% of adults with new-onset T1D are initially diagnosed with T2D.² Most commonly, clinical phenotype alone is used to classify T1D and T2D in adults. Unless there is an atypical feature of the clinical presentation, other classification tools, such as T1D autoantibodies or C-peptide, are not routinely measured at diagnosis in adults.³ Because individuals with T1D and T2D may have overlapping characteristics, these classification tools may not be sufficiently utilized to make an accurate diagnosis.^2,4,5 A comprehensive tool which integrates various parameters is needed to distinguish T1D and T2D in an adult with new onset of diabetes because correct classification can lead to appropriate therapies and avoid serious complications like diabetic ketoacidosis (DKA).

Existing classification tools tend to generally divide input variables or test results into binary (presence/absence) or quantitative dichotomous outcomes with a cutpoint.^6,7 For example, 10% to 15% of people with T1D do not have autoantibody seropositivity.^8-10 Therefore, if autoantibody status was used as a definitive diagnosis for T1D, then T1D would be misclassified. Likewise, if a phenotypic characteristic such as body mass index (BMI) less than 30 is used as a cut-off to distinguish T1D and T2D, then people with T1D and obesity may be misclassified as T2D. Each of these classification tools are often used in isolation, despite every person having many distinguishing phenotypic, biomarker (autoantibodies and C-peptide), and genotypic tests which can inform diagnosis. Rather than a binary outcome when these classification tools are applied, a continuous measurement based on variables encompassing phenotype, biomarkers, and genotype could be used to estimate probability of T1D versus T2D and assist clinicians in making a correct diagnosis at diabetes onset. By creating a flowchart with multiple branch points for each classification feature at each decision point and the likelihoods of T1D and T2D for each branch, a more accurate and useful decision tree for classifying types of diabetes can be created. Therefore, an ideal classification tool should not only be able to incorporate more than two possible decisions for a given classification feature but should also accommodate the possibility that for a given outcome branch based on one or more input variables, there could be a specific likelihood of T1D and a specific likelihood of T2D.

A nuanced probable classification tool is now needed in adults with new-onset diabetes to avoid diabetes type misclassification. This tool would provide the likelihood of a diagnosis type in order to avoid (1) misdiagnosis of actual T1D as T2D, (2) misdiagnosis of actual T2D or monogenic diabetes as T1D, and (3) misclassified patients in future population health studies which will lead to incorrect conclusions and suboptimal patient outcomes. Here we describe in detail the reasons for and potential harm from diabetes misclassification in order to support the development of a comprehensive diabetes classification tool.

Types of Diabetes

It is important to correctly diagnose T1D in adults based on multiple criteria and not automatically attribute the age of onset and BMI as the only distinguishing features. Because the pathophysiology is different for these two diseases, the proper treatments are also different.

Misclassification of new-onset diabetes in adults could adversely impact glucose management, resulting in incorrect care and suboptimal outcomes and research populations (resulting in incorrect policies). The misclassification of T1D in a patient with T2D resulting in unnecessary insulin treatment can be associated with significant emotional distress in individuals with diabetes and their families.¹¹ Conversely, the misdiagnosis of T2D in a patient with T1D might put the patient at risk of developing progressive hyperglycemia or even DKA while receiving treatment(s) other than insulin. Providing an ineffective treatment for a patient because of diabetes type misclassification could lead to a loss of confidence in modern effective treatments for diabetes.

Among adults with new-onset T1D, 40% are initially misclassified, with a higher likelihood of misdiagnosis as age increases.² More new-onset cases of T1D occur in adults than in children in the US.² However, given the much greater number of new-onset cases of T2D in adults than in children, adult-onset cases of T1D represent a smaller proportion of new adult-onset cases of diabetes compared with the proportion of children with newly diagnosed diabetes who are diagnosed with T1D. Five incorrect assumptions contribute to frequent mistaken diagnoses of T2D in patients with adult-onset diabetes who actually have T1D. These assumptions include that (1) any adult who develops diabetes must automatically have a diagnosis of T2D (even though T1D can arise in adults),¹² (2) pancreatic autoantibody tests, which can support a diagnosis of T1D, are frequently not needed or requested (even though they can be used to establish a diagnosis of T1D),¹³ (3) the negative predictive value of pancreatic autoantibody tests is poor, because a proportion of people with T1D at diagnosis are autoantibody negative,¹⁴ (4) if insulin therapy is not required at the time of a diabetes diagnosis, then the diagnosis must be T2D (even though T1D can be the cause, such as in cases of latent autoimmune diabetes of adults),¹⁵ and (5) the presence of obesity or metabolic syndrome automatically indicates a diagnosis of T2D; these traits are common in the general population and in fact do not rule out a diagnosis of T1D.¹⁶ Conversely, an incorrect assumption may contribute to mistaken diagnoses of T1D in patients with adult-onset diabetes who actually have T2D. This type of assumption relates to the expectation that low body weight, severe hyperglycemia, and even ketoacidosis must indicate a diagnosis of T1D (even though these symptoms can also occur in T2D).¹⁷

“Double diabetes,” while not explicitly defined in the American Diabetes Association’s (ADA’s) Diagnosis and Classification of Diabetes: Standards of Care in Diabetes—2025,¹⁸ is a term used for a condition where features of both T1D and T2D coexist in an individual.¹⁹ (The ADA does acknowledge that individuals may be diagnosed with T1D, while also exhibiting symptoms of T2D.)¹⁸ This disorder may refer to a person with T1D who develops insulin resistance, which is a typical feature of T2D, or else a person with T2D who produces autoantibodies against pancreatic beta cells, which is a characteristic feature of T1D.²⁰ Latent autoimmune diabetes in adults (LADA) is sometimes referred to as Type 1.5 diabetes²¹ and accounts for 5% to 13% of patients with T2D.²² This type of diabetes identifies a group with clinical and genetic features intermediate between typical T1D and T2D.²² Ketosis-prone diabetes (KPD) is a distinct type of diabetes that presents with reversible beta cell dysfunction in patients who are not insulin dependent. This type of diabetes is neither type 1 nor type 2.^20,23 Four different subtypes of KPD can be distinguished based on the presence or absence of beta cell autoantibodies and beta cell functional reserve.²⁴

T1D Misdiagnosed as T2D

Three explanations for T1D being mistakenly diagnosed as T2D can include (1) T-lymphocyte activation in the absence of circulating islet cell antibodies, (2) fluctuating insulin requirements, and (3) other unknown causes. T1D may be misdiagnosed as T2D when detectable islet autoantibodies (IAb) are not present, leading to incorrect treatments incorrectly focused on T2D. Up to 15% of individuals with a clinical phenotype of T1D do not have evidence of pancreatic islet autoantibodies.²⁵ In patients with islet autoantibody negative (IAb−) T1D, the diabetes may still have an immune-related cause, even though autoantibodies are not present.²⁶ The actual destruction of pancreatic beta cells is primarily mediated by T cells.²⁶ In some individuals with T1D, immune cell infiltration within islets (particularly by CD4+ and CD8+ T cells) is observed even when autoantibodies may not be detectable, which indicates an ongoing immune-mediated processes. Therefore, the presence of circulating autoantibodies is not always necessary for a diagnosis of T1D.²⁷ Another potential reason for IAb− T1D is defects in antibody production. People with phenotypic T1D who are not IAb+ may still have T1D²⁵ because of defects in antibody production. Patients with common variable immunodeficiency disease (CVID) may develop T1D but be unable to produce antibodies.²⁸ CVID can be detected through total antibody testing, and this information may assist clinicians in correctly diagnosing IAb− T1D.²⁵

In addition, a certain subset of patients with IAb− T1D and fluctuating insulin requirements may be misdiagnosed because their phenotype has features similar to the classic phenotypes of both T1D and T2D. These patients may have transiently present autoantibodies²⁹ or their autoantibodies may become undetectable over time.³⁰ These patients generally present with DKA, high HbA1c, and/or are not obese—hallmarks of T1D—yet also attain T1D remission with monotherapy or combination therapy, weight loss, and/or carbohydrate restriction.³¹ Nonfasting C-peptide values for this group may range from 200 to 700 pmol/L,³¹ higher than that of patients with IAb+ T1D. Patel et al³¹ terms this group “autoantibody-negative T1D in remission (ANG1R),” and proposes that individuals in this group have changeable beta-cell function. This form of IAb− T1D is particularly common in patients of Hispanic, African American, and Asian descent.³² Those with “ANG1R” have a similar phenotype to those with KPD.³¹

Finally, it is possible that in individuals with IAb− T1D, there are yet undiscovered autoantibodies, beyond GAD, IA2, ZnT8, or IAA,²⁵ or autoantibodies present at concentrations too low to be detected. For instance, the peak incidence of IAbs (per 1000 person-years) occurs from 9 months to 2 years of age and may decrease afterward.³³ In both of these cases, any autoantibodies that are present may become undetectable by current autoantibody assays. It may be particularly useful to study novel autoantibodies in demographics more likely to have IAb− T1D, such as patients with Asian or African ethnicity.³²

T2D or Monogenic Diabetes Misdiagnosed as T1D

Two explanations for different forms of diabetes being mistakenly diagnosed as T1D can include monogenic diabetes and other diagnostic uncertainties. Patients who are diagnosed with IAb− T1D may not have T1D at all, and instead, monogenic diabetes or T2D. Both cases of misdiagnosis can lead to suboptimal treatment decisions and outcomes for the patient.

Monogenic diabetes is neither T1D nor T2D. It is a distinct form of diabetes caused by mutations in a single gene.³⁴ This type of diabetes accounts for approximately 2% of patients with diabetes.³⁵ Monogenic diabetes results from mutations in genes that are crucial for beta cell development or function.³² The classic phenotype of monogenic diabetes includes age of onset less than 25 years, lean BMI, nonfasting C-peptide values of 200 to 700 pmol/L (within five hours of eating), and negative autoantibodies.³¹ Monogenic diabetes may be misdiagnosed as T1D at onset because of the overlapping phenotypic features between the two conditions. The most common form of monogenic diabetes is Maturity-onset diabetes of the young (MODY). MODY, itself, can be caused by mutations in at least 14 different genes, resulting in variable phenotypes for patients with MODY. For instance, individuals with MODY may have variable age of onset, symptoms, and response to treatment.³⁶ In addition to MODY, another form of monogenic diabetes is mitochondrial diabetes, which may present with a phenotype similar to that of either T1D or T2D, resulting in misdiagnosis by providers.³⁷ Mitochondrial diabetes can be differentiated from other forms of monogenic diabetes because it can present with symptoms related to multiple organs, is often transmitted maternally, and has an average age of onset of about 35 years old. The most common form of mitochondrial diabetes is maternally inherited diabetes and deafness (MIDD).³⁷

In addition to monogenic diabetes being misdiagnosed as T1D, it is possible for individuals with T2D to be misdiagnosed as having T1D. This may occur when an individual has certain phenotypic characteristics present in both T1D and T2D, and therefore some diagnostic uncertainty. For instance, individuals with A−B+ KPD³⁸ (A- to signify no documented IAb, and B+ to signify functional beta cell reserve) have overlapping phenotypic characteristics between that of T1D and that of T2D.³¹ This phenotype of A−B+ KPD resembles that of a form of T1D (ANG1R), and differences between KPD and ANG1R are presented in Table 1.

Table 1.

Comparison of Autoantibody-Negative T1D in Remission (ANG1R) and A−B+ Ketosis-Prone Diabetes (KPD).^{25,29-32,38-41}

	ANG1R	A−B+ KPD
More similar to T1D or T2D?	Considered a form of T1D	Considered a form of T2D
Beta cell features	Progressive loss of beta cell function, mainly due to insulin resistance and partially due to immune-mediated action	Reversible beta cell dysfunction, leading to insulin resistance. Individuals may also develop beta cell failure and insulin dependence due to unknown reasons
Possible remission?	Remission may occur with single drugs³¹ or combination therapy (defined as two or more antidiabetic drugs administered simultaneously),⁴⁰ weight loss, and/or carbohydrate restriction.	Potential for prolonged remission with or without medication
Autoantibody status	Individuals generally have autoantibodies, but they may be fluctuating or undetectable.	No autoantibodies detected
Age at diagnosis	Older age at diagnosis	Older age at diagnosis
Ethnicity	More common in patients of Hispanic, African American, and Asian descent	More common in African American and Hispanic populations, with a lower prevalence in Asian populations^39,41
Phenotype	Lean or overweight, present with DKA, high HbA1c, nonfasting C-peptide between 200 and 700 pmol/L	Overweight or obese, functional beta cell reserve, variable C-peptide levels
Presence of DKA	Less common than in individuals with autoantibody positive T1D	Commonly present with DKA precipitated by illness, may be insulin independent after treatment of acute hyperglycemia

Abbreviations: ANG1R, autoantibody-negative T1D in remission; A−B+ KPD, autoantibody-negative beta-cell function positive ketosis-prone diabetes; DKA, diabetic ketoacidosis; HbA1c, hemoglobin A1c; T1D, type 1 diabetes; T2D, type 2 diabetes.

Diagnostic Tools for Classification

A tool for classifying T1D and T2D in adult-onset diabetes is needed to aid correct diagnoses. The most widely used diagnostic tests for this purpose include (1) phenotypic differences between T1D and T2D (such as age of onset, BMI, tests of insulin sensitivity, and history of ketoacidosis), (2) pancreatic autoantibodies, (3) insulin secretion manifested by measuring C-peptide, and (4) genetic risk scores (although not typically used to differentiate between T1D and T2D, this testing can be used to support identification of monogenic forms of diabetes). Combining these types of tests can lead to a likely diagnosis.^42,43 Each individual test is not diagnostic in itself, but a combination of results of these four classifications can lead to a likely diagnosis. The goal of a tool development would be to develop a practical flowchart for classifying the type of diabetes for an adult with decision points connected by arrows. There would be likelihoods at the decision points using both binary and quantitative classification with cut points to determine an overall percentage of likelihood of T1D and T2D for each combination of characteristics. The percentages will be based on a consensus of experts and databases of patients with clearly defined features of T1D and T2D as well as articles in the medical literature. To our knowledge, this type of flowchart with percentage likelihoods for combinations of features or biomarkers to classify new-onset diabetes patients as T1D or T2D has not previously been reported. We will also develop an accompanying software program to classify individual patients. This flowchart could be used in an electronic health record to classify patients and assign treatment based on the classification. In this tool, we plan to consider differences in diabetes diagnosis and classification found in different races/ethnicities. A generic example of this type of flowchart is illustrated in Figure 1. More specifically, we will be able to use existing databases of patients and existing literature to determine which percentages of patients with a clinical diagnosis of T1D and T2D exhibit each specific characteristic (e.g. autoantibody status, C-peptide), and then apply this information to create a tool to classify patients with adult-onset diabetes.

Figure 1.

Example of a generic flowchart with percentage likelihoods for combinations of features or biomarkers that can be used to help diagnose adults with new-onset diabetes. Abbreviations: T1D, type 1 diabetes; T2D, type 2 diabetes.

The 2025 ADA Standards of Care in Diabetes presents a flowchart for investigation of suspected T1D in newly diagnosed adults, based on data from White European populations.¹⁸ The figure legend stipulates that no single clinical feature confirms T1D in isolation. This figure contains two branches for each diagnostic test (except C-peptide release, which has three different branches). It is likely that development of a more detailed flowchart with more branches for each diagnostic test and with likelihoods of T1D versus T2D for each branch point, compared with the ADA flowchart, could provide an even more nuanced classification using these four tests.

An emerging tool to diagnose different forms of diabetes, coming out of recent research in the area, is the use of machine learning algorithms.⁴⁴ When diagnosing diabetes, it is important to consider demographic factors in addition to IAb status, since the characteristics of T1D may differ in different populations, and machine learning may make it much more accessible to consider these demographic factors all at once.⁴⁵ Using machine learning applied to metabolic subphenotypes from glucose time series, Metwally et al⁴⁶ demonstrated that individuals with normoglycemia and prediabetes exhibit heterogeneity in four distinct physiologic processes that contribute to disordered glucose metabolism, including muscle insulin resistance, beta cell dysfunction, impaired incretin effect, and hepatic insulin resistance. In addition, Tan et al⁴⁷ recently developed machine learning algorithms to diagnose T1D using either the gut microbiome, serum metabolite, or serum lipid signatures. Machine learning using metabolite composition diagnosed T1D with a fairly high accuracy of 0.933.

However, these unsupervised approaches will only be accurate if developed using accurate data about patient cohorts with T1D, T2D, and other forms of diabetes. Similarly, researchers and clinicians will only be able to determine new ways to diagnose and treat different forms of diabetes if their research cohorts are accurately defined by population. Hence, it is crucial to correctly distinguish the different forms of diabetes not only to support the treatment of individuals, but also to support research at a population level.

Conclusions

A flowchart tool for classifying the type of diabetes for an adult with decision points connected by arrows and likelihoods at the decision points, along with development of an accompanying software program to classify individual patients, will be a practical diagnostic tool with favorable therapeutic implications. With new treatments becoming available for T1D and T2D, it is important to make a correct diagnosis so that effective treatment can be prescribed.

Footnotes

Abbreviations

A−B+ KPD, autoantibody-negative beta-cell function positive ketosis-prone diabetes; ADA, American Diabetes Association; BMI, body mass index; CVID, common variable immunodeficiency disease; diabetic ketoacidosis, DKA; HbA1c, hemoglobin A1c; IAb, islet autoantibody; IAb−, islet autoantibody negative; KPD, ketosis-prone diabetes; LADA, latent autoimmune diabetes in adults; MIDD, maternally inherited diabetes and deafness; MODY, maturity-onset diabetes of the young; T1D, type 1 diabetes; T2D, type 2 diabetes.

Declaration of Conflicting Interests

The author(s) declared the following potential conflicts of interest with respect to the research, authorship, and/or publication of this article: A.T.A is a consultant for Liom.

C.N.H. is a consultant for Liom.

L.K.B has received consulting honoraria from Novo Nordisk, Endogenex, Sanofi, Dexcom, Bayer, Xeris, and Pfizer.

S.M. has received speaker honoraria from Lilly UK and Sanofi.

D.C.K. is a consultant for Afon, embecta, Glucotrack, Lifecare, Novo, Samsung, SynchNeuro, and Thirdwayv.

Funding

The author(s) received no financial support for the research, authorship, and/or publication of this article.

ORCID iDs

Alessandra T. Ayers

Cindy N. Ho

Liana K. Billings

Shivani Misra

David C. Klonoff

References

CDC. Diabetes. Type 2 diabetes. Published September 25, 2024. Accessed February 6, 2025. https://www.cdc.gov/diabetes/about/about-type-2-diabetes.html

Leslie

Evans-Molina

Freund-Brown

, et al. Adult-onset type 1 diabetes: current understanding and challenges. Diabetes Care. 2021;44(11):2449-2456. doi:10.2337/dc21-0770.

Mohan

Deepa

Somannavar

Datta

A simplified Indian diabetes risk score for screening for undiagnosed diabetic subjects. J Assoc Physicians India. 2005;53:759-763.

Muñoz

Floreen

Garey

, et al. Misdiagnosis and diabetic ketoacidosis at diagnosis of type 1 diabetes: patient and caregiver perspectives. Clin Diabetes. 2019;37(3):276-281. doi:10.2337/cd18-0088.

Thomas

Lynam

Hill

, et al. Type 1 diabetes defined by severe insulin deficiency occurs after 30 years of age and is commonly treated as type 2 diabetes. Diabetologia. 2019;62(7):1167-1172. doi:10.1007/s00125-019-4863-8.

Holt

RIG

DeVries

Hess-Fischl

, et al. The management of type 1 diabetes in adults. A consensus report by the American Diabetes Association (ADA) and the European Association for the Study of Diabetes (EASD). Diabetes Care. 2021;44(11):2589-2625. doi:10.2337/dci21-0043.

Hörber

Achenbach

Schleicher

Peter

Harmonization of immunoassays for biomarkers in diabetes mellitus. Biotechnol Adv. 2020;39:107359. doi:10.1016/j.biotechadv.2019.02.015.

Imagawa

Hanafusa

Miyagawa

Matsuzawa

A novel subtype of type 1 diabetes mellitus characterized by a rapid onset and an absence of diabetes-related antibodies. N Engl J Med. 2000;342(5):301-307. doi:10.1056/NEJM200002033420501.

American Diabetes Association. Diagnosis and classification of diabetes mellitus. Diabetes Care. 2014;37(suppl 1):S81-S90. doi:10.2337/dc14-S081.

10.

Bravis

Kaur

Walkey

, et al. Relationship between islet autoantibody status and the clinical characteristics of children and adults with incident type 1 diabetes in a UK cohort. BMJ Open. 2018;8(4):e020904. doi:10.1136/bmjopen-2017-020904.

11.

Tosur

Huang

Inglis

Aguirre

Redondo

MJ.

Inaccurate diagnosis of diabetes type in youth: prevalence, characteristics, and implications. Sci Rep. 2024;14(1):8876. doi:10.1038/s41598-024-58927-6.

12.

Vieira

. Misdiagnosed with type 2 diabetes: why adults should be screened for autoantibodies. T1D Exchange. Published November 21, 2023. Accessed January 8, 2025. https://t1dexchange.org/adults-misdiagnosed-with-type-2-diabetes/

13.

Manov

Chauhan

Dhillon

Donepudi

Unmasking type 1 diabetes in adults: insights from two cases revealing misdiagnosis as type 2 diabetes, with emphasis on autoimmunity and continuous glucose monitoring. Cureus. 2023;15(7):e42459. doi:10.7759/cureus.42459.

14.

Thomas

Jones

AG.

The challenges of identifying and studying type 1 diabetes in adults. Diabetologia. 2023;66(12):2200-2212. doi:10.1007/s00125-023-06004-4.

15.

Buzzetti

Tuomi

Mauricio

, et al. Management of latent autoimmune diabetes in adults: a consensus statement from an international expert panel. Diabetes. 2020;69(10):2037-2047. doi:10.2337/dbi20-0017.

16.

Krause

De Vito

Type 1 and type 2 diabetes mellitus: commonalities, differences and the importance of exercise and nutrition. Nutrients. 2023;15(19):4279. doi:10.3390/nu15194279.

17.

Umpierrez

Davis

ElSayed

, et al. Hyperglycemic crises in adults with diabetes: a consensus report. Diabetes Care. 2024;47(8):1257-1275. doi:10.2337/dci24-0032.

18.

American Diabetes Association Professional Practice Committee. 2. Diagnosis and classification of diabetes: standards of care in diabetes-2025. Diabetes Care. 2025;48(suppl 1):S27-S49. doi:10.2337/dc25-S002.

19.

Bielka

Przezak

Molęda

Pius-Sadowska

Machaliński

Double diabetes-when type 1 diabetes meets type 2 diabetes: definition, pathogenesis and recognition. Cardiovasc Diabetol. 2024;23(1):62. doi:10.1186/s12933-024-02145-x.

20.

Redondo

Morgan

NG.

Heterogeneity and endotypes in type 1 diabetes mellitus. Nat Rev Endocrinol. 2023;19(9):542-554. doi:10.1038/s41574-023-00853-0.

21.

Cleveland Clinic. What is LADA diabetes? Accessed January 8, 2025. https://my.clevelandclinic.org/health/diseases/lada-diabetes

22.

Stone

Balasubramanyam

Posey

JE.

Atypical diabetes: what have we learned and what does the future hold?

Diabetes Care. 2024;47(5):770-781. doi:10.2337/dci23-0038.

23.

Makahleh

Othman

Vedantam

Ketosis-prone type 2 diabetes mellitus: an unusual presentation. Cureus. 2022;14(10):e30031. doi:10.7759/cureus.30031.

24.

Kubota-Mishra

Huang

Minard

, et al. High prevalence of A-β+ ketosis-prone diabetes in children with type 2 diabetes and diabetic ketoacidosis at diagnosis: evidence from the rare and atypical diabetes network (RADIANT). Pediatr Diabetes. 2024;2024:5907924. doi:10.1155/2024/5907924.

25.

Patel

Fourlanos

Greenfield

. Autoantibody-negative type 1 diabetes: a neglected subtype. Trends Endocrinol Metab. 2021;32(5):295-305. doi:10.1016/j.tem.2021.02.001.

26.

Burrack

Martinov

Fife

BT.

T cell-mediated beta cell destruction: autoimmunity and alloimmunity in the context of type 1 diabetes. Front Endocrinol (Lausanne). 2017;8:343. doi:10.3389/fendo.2017.00343.

27.

Pihoker

Gilliam

Hampe

Lernmark

Autoantibodies in diabetes. Diabetes. 2005;54(suppl 2):S52-S61. doi:10.2337/diabetes.54.suppl_2.s52.

28.

Milota

Šumník

Obermannová

, et al. Negativity for specific autoantibodies in patients with type 1 diabetes that developed on a background of common variable immunodeficiency. Int Arch Allergy Immunol. 2015;168(3):197-204. doi:10.1159/000441723.

29.

Jia

Effective assay technologies fit for large-scale population screening of type 1 diabetes. Front Clin Diabetes Healthc. 2022;3:1034698. doi:10.3389/fcdhc.2022.1034698.

30.

Niskanen

Tuomi

Karjalainen

Groop

Uusitupa

MI.

GAD antibodies in NIDDM. Ten-year follow-up from the diagnosis. Diabetes Care. 1995;18(12):1557-1565. doi:10.2337/diacare.18.12.1557.

31.

Patel

Fourlanos

Greenfield

. Classification of type 1 diabetes: a pathogenic and treatment-based classification. Diabetes Metab Syndr. 2024;18(3):102986. doi:10.1016/j.dsx.2024.102986.

32.

Abdel-Karim

Haris

Afyouni

, et al. The epidemiology and genetic analysis of children with idiopathic type 1 diabetes in the state of Qatar. J Endocr Soc. 2021;5(10):bvab131. doi:10.1210/jendso/bvab131.

33.

Ziegler

Bonifacio

; BABYDIAB-BABYDIET Study Group. Age-related islet autoantibody incidence in offspring of patients with type 1 diabetes. Diabetologia. 2012;55(7):1937-1943. doi:10.1007/s00125-012-2472-x.

34.

Zhang

Colclough

Gloyn

Pollin

TI.

Monogenic diabetes: a gateway to precision medicine in diabetes. J Clin Invest. 2021;131(3):e142244. doi:10.1172/JCI142244.

35.

Harris

Letourneau

Greeley

SAW

. Monogenic diabetes: the impact of making the right diagnosis. Curr Opin Pediatr. 2018;30(4):558-567. doi:10.1097/MOP.0000000000000643.

36.

Hoffman

Fox

Anastasopoulou

Jialal

Maturity Onset Diabetes in the Young. Treasure Island, FL: StatPearls Publishing; 2023. Accessed October 8, 2024. https://www.ncbi.nlm.nih.gov/books/NBK532900/

37.

Yeung

Al Jundi

Gubbi

, et al. Management of mitochondrial diabetes in the era of novel therapies. J Diabetes Complications. 2021;35(1):107584. doi:10.1016/j.jdiacomp.2020.107584.

38.

Balasubramanyam

Defining and classifying new subgroups of diabetes. Annu Rev Med. 2021;72:63-74. doi:10.1146/annurev-med-050219-034524.

39.

Lebovitz

Banerji

MA.

Ketosis-prone diabetes (Flatbush diabetes): an emerging worldwide clinically important entity. Curr Diab Rep. 2018;18(11):120. doi:10.1007/s11892-018-1075-4.

40.

Xie

Hao

, et al. Benefits and risks of drug combination therapy for diabetes mellitus and its complications: a comprehensive review. Front Endocrinol (Lausanne). 2023;14:1301093. doi:10.3389/fendo.2023.1301093.

41.

Shankar

Chowdhury

Gousy

Parthasarathi

An insight into Flatbush diabetes: a rare form of diabetes. Cureus. 2022;14(1):e21567. doi:10.7759/cureus.21567.

42.

Lynam

McDonald

Hill

, et al. Development and validation of multivariable clinical diagnostic models to identify type 1 diabetes requiring rapid insulin therapy in adults aged 18-50 years. BMJ Open. 2019;9(9):e031586. doi:10.1136/bmjopen-2019-031586.

43.

Foteinopoulou

Clarke

CAL

Pattenden

, et al. Impact of routine clinic measurement of serum C-peptide in people with a clinician-diagnosis of type 1 diabetes. Diabet Med. 2021;38(7):e14449. doi:10.1111/dme.14449.

44.

Iparraguirre-Villanueva

Espinola-Linares

Flores Castañeda

Cabanillas-Carbonell

Application of machine learning models for early detection and accurate classification of type 2 diabetes. Diagn Basel Switz. 2023;13(14):2383. doi:10.3390/diagnostics13142383.

45.

Lugner

Rawshani

Helleryd

Eliasson

Identifying top ten predictors of type 2 diabetes through machine learning analysis of UK Biobank data. Sci Rep. 2024;14(1):2102. doi:10.1038/s41598-024-52023-5.

46.

Metwally

Perelman

Park

, et al. Prediction of metabolic subphenotypes of type 2 diabetes via continuous glucose monitoring and machine learning. Nat Biomed Eng. 2024. doi:10.1038/s41551-024-01311-6.

47.

Tan

Shi

Yue

, et al. Machine learning approach reveals microbiome, metabolome, and lipidome profiles in type 1 diabetes. J Adv Res. 2024;64:213-221. doi:10.1016/j.jare.2023.11.025.