Automated ICD Coding Using Deep Learning

Medical Coding ◽

Huge Data ◽

ICD (international classification of diseases) is a system developed by the WHO in which every unique diagnosis and procedure has a unique code. It provides a standardized way to represent medical information and makes it sharable and comparable across different hospitals and countries. Currently, the task of assigning ICD codes to patient discharge summaries is performed manually by medical coders. Manual coding is costly, time consuming, and inefficient for huge data. So, the healthcare industry requires automated solutions to make the medical coding more efficient, accurate, and consistent. In this study, the automated ICD-9 coding is approached as a multi-label text classification problem. A deep learning system is presented to assign ICD-9 codes automatically to the patient discharge summaries. Convolutional neural networks and word2vec model are combined to automatically extract features from the input text. The best model has achieved 83.28% accuracy. The results of this research prove the usability of deep learning for multi-label text classification and medical coding.

Converging Semantic Knowledge and Deep Learning for Medical Coding

International Journal of Privacy and Health Information Management ◽

10.4018/ijphim.2019070103 ◽

2019 ◽

Vol 7 (2) ◽

pp. 33-52

Author(s):

Nuria Garcia-Santa ◽

Beatriz San Miguel ◽

Takanori Ugai

Keyword(s):

Deep Learning ◽

Semantic Knowledge ◽

Clinical Notes ◽

Medical Coding ◽

Advantages And Disadvantages ◽

Medical Reports ◽

Mimic Iii

The field of medical coding enables to assign codes of medical classifications such as the international classification of diseases (ICD) to clinical notes, which are medical reports about patients' conditions written by healthcare professionals in natural language. These texts potentially include medical terms that define diagnosis, symptoms, drugs, treatments, etc., and the use of spontaneous language is challenging for automatic processing. Medical coding is usually performed manually by human medical coders becoming time-consuming and prone to errors. This research aims at developing new approaches that combine deep learning elements together with traditional technologies. A semantic-based proposal supported by a proprietary knowledge graph (KG), neural network implementations, and an ensemble model to resolve the medical coding are presented. A comparative discussion between the proposals where the advantages and disadvantages of each one is analysed. To evaluate approaches, two main corpus have been used: MIMIC-III and private de-identified clinical notes.

Proceedings of the Twenty-Ninth International Joint Conference on Artificial Intelligence ◽

Generalized Zero-Shot Text Classification for ICD Coding

10.24963/ijcai.2020/556 ◽

2020 ◽

Author(s):

Congzheng Song ◽

Shanghang Zhang ◽

Najmeh Sadoughi ◽

Pengtao Xie ◽

Eric Xing

Keyword(s):

Text Classification ◽

Classification Problem ◽

Fine Grained ◽

Clinical Document ◽

Previous State ◽

Label Distribution ◽

Generalized Zero ◽

Icd Coding

The International Classification of Diseases (ICD) is a list of classification codes for the diagnoses. Automatic ICD coding is a multi-label text classification problem with noisy clinical document inputs and long-tailed label distribution, making it difficult for fine-grained classification on both frequent and zero-shot codes at the same time, i.e. generalized zero-shot ICD coding. In this paper, we propose a latent feature generation framework to improve the prediction on unseen codes without compromising the performance on seen codes. Our framework generates semantically meaningful features for zero-shot codes by exploiting ICD code hierarchical structure and reconstructing the code-relevant keywords with a novel cycle architecture. To the best of our knowledge, this is the first adversarial generative model for generalized zero-shot learning on multi-label text classification. Extensive experiments demonstrate the effectiveness of our approach. On the public MIMIC-III dataset, our methods improve the F1 score from nearly 0 to 20.91% for the zero-shot codes, and increase the AUC score by 3% (absolute improvement) from previous state of the art. Code is available at https://github.com/csong27/gzsl_text.

693. Performance of ICD Code Versus Discharge Summary based Query for Endocarditis Cohort Identification

Open Forum Infectious Diseases ◽

10.1093/ofid/ofab466.890 ◽

2021 ◽

Vol 8 (Supplement_1) ◽

pp. S448-S448

Author(s):

H Nina Kim ◽

Ayushi Gupta ◽

Kristine F Lan ◽

Jenell C Stewart ◽

Shireesha Dhanireddy ◽

...

Keyword(s):

Infective Endocarditis ◽

Discharge Summary ◽

Admission Diagnosis ◽

Test Characteristics ◽

Clinical Narrative ◽

Icd 10 ◽

Icd Codes ◽

Abstract Background Studies on infective endocarditis (IE) have relied on International Classification of Diseases (ICD) codes to identify cases but few have validated this method which may be prone to misclassification. Examination of clinical narrative data could offer greater accuracy and richness. Methods We evaluated two algorithms for IE identification from 7/1/2015 to 7/31/2019: (1) a standard query of ICD codes for IE (ICD-9: 424.9, 424.91, 424.99, 421.0, 421.1, 421.9, 112.81, 036.42 and ICD-10: I38, I39, I33, I33.9, B37.6 and A39.51) with or without procedure codes for echocardiogram (93303-93356) and (2) a key word, pattern-based text query of discharge summaries (DS) that selected on the term “endocarditis” in fields headed by “Discharge Diagnosis” or “Admission Diagnosis” or similar. Further coding extracted the nature and type of valve and the organism responsible for the IE if present in DS. All identified cases were chart reviewed using pre-specified criteria for true IE. Positive predictive value (PPV) was calculated as the total number of verified cases over the algorithm-selected cases. Sensitivity was the total number of algorithm-matched cases over a final list of 166 independently identified true IE cases from ID and Cardiology services. Specificity was defined using 119 pre-adjudicated non-cases minus the number of algorithm-matched cases over 119. Results The ICD-based query identified 612 individuals from July 2015 to July 2019 who had a hospital billing code for infective endocarditis; of these, 534 also had an echocardiogram. The DS query identified 387 cases. PPV for the DS query was 84.5% (95% confidence interval [CI] 80.6%, 87.8%) compared with 72.4% (95% CI 68.7%, 75.8%) for ICD only and 75.8% (95% CI 72.0%, 79.3%) for ICD + echo queries. Sensitivity was 75.9% for the DS query and 86.8-93.4% for the ICD queries. Specificity was high for all queries >94%. The DS query also yielded valve data (prosthetic, tricuspid, pulmonic, aortic or mitral) in 60% and microbiologic data in 73% of identified cases with an accuracy of 94% and 90% respectively when assessed by chart review. Table 1. Test Characteristics of Three Electronic Health Record Queries for Infective Endocarditis Conclusion Compared to traditional ICD-based queries, text-based queries of discharge summaries have the potential to improve precision of IE case ascertainment and extract key clinical variables. Disclosures All Authors: No reported disclosures

Basic Principles and Benefits of Various Classification Systems in Health

Handbook of Research on Informatics in Healthcare and Biomedicine ◽

10.4018/978-1-59140-982-3.ch007 ◽

2011 ◽

pp. 51-58

Author(s):

Dimitra Petroudi ◽

Athanasios Zekios

Keyword(s):

Information Systems ◽

Health Professionals ◽

Classification Systems ◽

Coding System ◽

Basic Principles ◽

Medical Coding ◽

Coding Systems ◽

Classification Of Diseases

The introduction of information systems in health progressively led tï coding systems. The purposes of these systems are: recording causes of death, coding diseases and procedures, etc. The most important medical coding system in our days is ICD (International Classification of Diseases). Other coding systems that health professionals use are: SNOMED, LOINC, MeSH, UMLS, DSM, DRG and HCPCS. There are also many Nursing Classification Systems, such as: NANDA, NIC, NOC, ICNP, Omaha and HHCC. This chapter describes these coding systems and their advantages.

Text Summarization and Classification of Clinical Discharge Summaries using Deep Learning

10.36227/techrxiv.12059019.v1 ◽

2020 ◽

Author(s):

Alizar Marchawala ◽

Preetkumar Patel ◽

Khushal Paresh Thaker ◽

Hardik Gunjal ◽

Abhishek nagrecha ◽

...

Keyword(s):

Neural Networks ◽

Deep Learning ◽

Convolutional Neural Networks ◽

Diagnostic Procedure ◽

Patient Discharge ◽

Text Summarization ◽

Automated Classification ◽

Complex Features ◽

<p>This paper implements the automated classification of patient discharge notes into standard disease labels which includes the name of the diagnostic procedure required. In this approach, we use Convolutional Neural Networks to classify and represent complex features from the medical discharge summaries using the MT sample dataset. We make use of GloVE to have a pretrained model learn from it.<b></b></p>

Medical services provided on the ‘Harmonious Mission—2017’ Peace Ark from China

BMJ Military Health ◽

10.1136/bmjmilitary-2020-001659 ◽

2021 ◽

pp. e001659

Author(s):

Bihan Tang ◽

Y Han ◽

X Liu ◽

H Zhang ◽

M Li ◽

...

Keyword(s):

Low Income ◽

Medical Information ◽

Demographic Data ◽

Cataract Extraction ◽

Medical Personnel ◽

Low Income Countries ◽

Medical Services ◽

General Internal ◽

Classification Of Diseases

IntroductionThe Chinese Naval ship Peace Ark provided humanitarian medical services to people in eight low-income countries in Africa and Asia during the 2017 “Harmonious Mission’. The expedition lasted 155 days. Our study aimed to analyse the details of the medical services provided including outpatient care, medical patrol, operations, examinations and medications.MethodThe patient demographic data and medical information were extracted from electronic medical records. The diagnoses and procedures aboard were coded by the International Classification of Diseases, Ninth Revision, Clinical Modification (ICD-9-CM). The sociodemographic data of the medical staff aboard were collected via questionnaire. Descriptive statistics and statistical software (SAS, V.9.4) were used to analyse the data.ResultsIn total, 115 Chinese military medical personnel participated in the mission, completing a total of 50 758 outpatient visits, 10 232 medical patrols and 252 operations. The five most frequently used outpatient departments were ophthalmology, general surgery, general internal medicine, orthopaedics and traditional Chinese medicine. The five most common operations were lipoma excision, cataract extraction, skin tissue removal (such as warts and cysts), pterygium transposition and herniorrhaphy.ConclusionsOur study revealed the medical services in demand during the ‘Harmonious Mission—2017’. It is essential to report their experiences so that future ventures can provide medical services more effectively.

Text Summarization and Classification of Clinical Discharge Summaries using Deep Learning

10.36227/techrxiv.12059019 ◽

2020 ◽

Author(s):

Alizar Marchawala ◽

Preetkumar Patel ◽

Khushal Paresh Thaker ◽

Hardik Gunjal ◽

Abhishek nagrecha ◽

...

Keyword(s):

Neural Networks ◽

Deep Learning ◽

Convolutional Neural Networks ◽

Diagnostic Procedure ◽

Patient Discharge ◽

Text Summarization ◽

Automated Classification ◽

Complex Features ◽

A Deep Learning Framework for Automated ICD-10 Coding

Studies in Health Technology and Informatics - Public Health and Informatics ◽

10.3233/shti210178 ◽

2021 ◽

Author(s):

Abdelahad Chraibi ◽

David Delerue ◽

Julien Taillard ◽

Ismat Chaib Draa ◽

Régis Beuscart ◽

...

Keyword(s):

Deep Learning ◽

Language Processing ◽

Medical Information ◽

Coding System ◽

Learning Framework ◽

Diagnosis Codes ◽

Automated Coding ◽

International Statistical Classification

Icd 10 ◽

The International Statistical Classification of Diseases and Related Health Problems (ICD) is one of the widely used classification system for diagnoses and procedures to assign diagnosis codes to Electronic Health Record (EHR) associated with a patient’s stay. The aim of this paper is to propose an automated coding system to assist physicians in the assignment of ICD codes to EHR. For this purpose, we created a pipeline of Natural Language Processing (NLP) and Deep Learning (DL) models able to extract the useful information from French medical texts and to perform classification. After the evaluation phase, our approach was able to predict 346 diagnosis codes from heterogeneous medical units with an accuracy average of 83%. Our results were finally validated by physicians of the Medical Information Department (MID) in charge of coding hospital stays.

Child maltreatment surveillance following the ICD-10-CM transition, 2016-2018

Injury Prevention ◽

10.1136/injuryprev-2019-043579 ◽

2020 ◽

pp. injuryprev-2019-043579

Author(s):

Amy A Hunter ◽

Nina Livingston ◽

Susan DiVietro ◽

Laura Schwab Reese ◽

Kathryn Bentivegna ◽

...

Keyword(s):

Child Maltreatment ◽

Physical Abuse ◽

Delayed Diagnosis ◽

Secondary Data ◽

Medical Coding ◽

Icd 10 ◽

Ed Visits ◽

Maltreatment Type

BackgroundChild maltreatment is poorly documented in clinical data. The International Classification of Diseases and Related Health Problems, 10th Revision, Clinical Modification (ICD-10-CM) represents the first time that confirmed and suspected child maltreatment can be distinguished in medical coding. The utility of this distinction in practice remains unknown. This study aims to evaluate the application of these codes by patient demographic characteristics and injury type.MethodsWe conducted secondary data analysis of emergency department (ED) discharge records of children under 18 years with an ICD-10-CM code for confirmed (T74) or suspected (T76) child maltreatment. Child age, sex, race/ethnicity, insurance status and co-occurring injuries (S00-T88) were compared by maltreatment type (confirmed or suspected).ResultsFrom 2016 to 2018, child maltreatment was documented in 1650 unique ED visits, or 21.7 per 10 000 child ED visits. Suspected maltreatment was documented most frequently (58%). Half of all maltreatment-related visits involved sexual abuse, most often in females and individuals of non-Hispanic white race. Physical abuse was coded in 36% of visits; injuries to the head were predominant. Non-Hispanic black children were more frequently documented with confirmed physical abuse than suspected (38.7% vs 23.7%, p<0.01). The rate of co-occurring injuries documented with confirmed and suspected maltreatment differed by 30% (9.2 vs 12.5 per 10 000 ED visits, respectively).ConclusionsThe ability to discriminate confirmed and suspected maltreatment may help mitigate clinical barriers to maltreatment surveillance associated with delayed diagnosis and subsequent intervention. Racial disparities in suspected and confirmed cases were identified which may indicate biased diagnostic behaviours in the ED.

Auditing the Accuracy of Medical Diagnostic Coding based on International Classification of Diseases, Tenth Revision

10.32592/ircmj.2020.22.9.99 ◽

2020 ◽

Keyword(s):

Medical Information ◽

International Classification ◽

Cross Sectional ◽

Medical Diagnoses ◽

Diagnostic Coding ◽

Icd 10 ◽

Medical Diagnostic

Background: Medical diagnostic coding is used for the ease of retrieval and accuracy of medical information classification in health information systems. This information is the main source of decision making for health managers and policymakers in planning, epidemiological, and medical research at different levels. Objectives: The present study aimed to audit the accuracy of the ICD-10 (International Classification of Diseases, Tenth Revision) medical diagnosis code. Materials and Methods: The present cross-sectional study was performed on a sample of 692 hospitalized cases in 9 educational centers affiliated to Shahid Beheshti University of Medical Sciences in the first half of 2020. The content validity of the checklist was determined in this study, and the obtained data were analyzed in SPSS software using descriptive statistics. Results: The average accuracy of coding for the main medical diagnoses across all subjects was 70%, signifying that 30% of medical records contain coding errors. The highest and least accuracy values of diagnostic coding were 80% and 47%, respectively. The application of standard abbreviations and file legibility were recognized as variables affecting code accuracy. The highest precision percentage of codes attributed to other medical diagnoses, including ICD-10-based comorbidity and complication, was in 84%-85% of the participants. Conclusion: Given the importance of all-encompassing coding in retrieving medical information, research, and macro-health policymaking, the coding accuracy audit must be conducted on a regular basis. The interaction between coders and healthcare providers, coders' training, and improving the documentation process exerts a significant impact on the enhancement of coding accuracy.