scholarly journals Human- Versus Machine Learning–Based Triage Using Digitalized Patient Histories in Primary Care: Comparative Study (Preprint)

2020 ◽  
Author(s):  
Artin Entezarjou ◽  
Anna-Karin Edstedt Bonamy ◽  
Simon Benjaminsson ◽  
Pawel Herman ◽  
Patrik Midlöv
Keyword(s):  
Machine Learning ◽  
Primary Care ◽  
Physical Examination ◽  
Medical History ◽  
Primary Care Physicians ◽  
Interrater Reliability ◽  
Reference Group ◽  
Majority Vote ◽  
Medical Histories ◽  
Using Data

BACKGROUND Smartphones have made it possible for patients to digitally report symptoms before physical primary care visits. Using machine learning (ML), these data offer an opportunity to support decisions about the appropriate level of care (triage). OBJECTIVE The purpose of this study was to explore the interrater reliability between human physicians and an automated ML-based triage method. METHODS After testing several models, a naïve Bayes triage model was created using data from digital medical histories, capable of classifying digital medical history reports as either in need of urgent physical examination or not in need of urgent physical examination. The model was tested on 300 digital medical history reports and classification was compared with the majority vote of an expert panel of 5 primary care physicians (PCPs). Reliability between raters was measured using both Cohen κ (adjusted for chance agreement) and percentage agreement (not adjusted for chance agreement). RESULTS Interrater reliability as measured by Cohen κ was 0.17 when comparing the majority vote of the reference group with the model. Agreement was 74% (138/186) for cases judged not in need of urgent physical examination and 42% (38/90) for cases judged to be in need of urgent physical examination. No specific features linked to the model’s triage decision could be identified. Between physicians within the panel, Cohen κ was 0.2. Intrarater reliability when 1 physician retriaged 50 reports resulted in Cohen κ of 0.55. CONCLUSIONS Low interrater and intrarater agreement in triage decisions among PCPs limits the possibility to use human decisions as a reference for ML to automate triage in primary care.

scholarly journals Human- Versus Machine Learning–Based Triage Using Digitalized Patient Histories in Primary Care: Comparative Study

10.2196/18930 ◽  
2020 ◽  
Vol 8 (9) ◽  
pp. e18930
Author(s):  
Artin Entezarjou ◽  
Anna-Karin Edstedt Bonamy ◽  
Simon Benjaminsson ◽  
Pawel Herman ◽  
Patrik Midlöv
Keyword(s):  
Machine Learning ◽  
Primary Care ◽  
Physical Examination ◽  
Medical History ◽  
Primary Care Physicians ◽  
Interrater Reliability ◽  
Reference Group ◽  
Majority Vote ◽  
Medical Histories ◽  
Using Data

Background Smartphones have made it possible for patients to digitally report symptoms before physical primary care visits. Using machine learning (ML), these data offer an opportunity to support decisions about the appropriate level of care (triage). Objective The purpose of this study was to explore the interrater reliability between human physicians and an automated ML-based triage method. Methods After testing several models, a naïve Bayes triage model was created using data from digital medical histories, capable of classifying digital medical history reports as either in need of urgent physical examination or not in need of urgent physical examination. The model was tested on 300 digital medical history reports and classification was compared with the majority vote of an expert panel of 5 primary care physicians (PCPs). Reliability between raters was measured using both Cohen κ (adjusted for chance agreement) and percentage agreement (not adjusted for chance agreement). Results Interrater reliability as measured by Cohen κ was 0.17 when comparing the majority vote of the reference group with the model. Agreement was 74% (138/186) for cases judged not in need of urgent physical examination and 42% (38/90) for cases judged to be in need of urgent physical examination. No specific features linked to the model’s triage decision could be identified. Between physicians within the panel, Cohen κ was 0.2. Intrarater reliability when 1 physician retriaged 50 reports resulted in Cohen κ of 0.55. Conclusions Low interrater and intrarater agreement in triage decisions among PCPs limits the possibility to use human decisions as a reference for ML to automate triage in primary care.

scholarly journals Machine Learning Models for Abnormality Detection in Musculoskeletal Radiographs

Reports ◽  
2019 ◽  
Vol 2 (4) ◽  
pp. 26 ◽  
Author(s):  
Govind Chada
Keyword(s):  
Machine Learning ◽  
Primary Care ◽  
Deep Learning ◽  
Transfer Learning ◽  
Primary Care Physicians ◽  
Screening Tools ◽  
Learning Approaches ◽  
Limited Data ◽  
Learning Models ◽  
High Recognition Accuracy

Increasing radiologist workloads and increasing primary care radiology services make it relevant to explore the use of artificial intelligence (AI) and particularly deep learning to provide diagnostic assistance to radiologists and primary care physicians in improving the quality of patient care. This study investigates new model architectures and deep transfer learning to improve the performance in detecting abnormalities of upper extremities while training with limited data. DenseNet-169, DenseNet-201, and InceptionResNetV2 deep learning models were implemented and evaluated on the humerus and finger radiographs from MURA, a large public dataset of musculoskeletal radiographs. These architectures were selected because of their high recognition accuracy in a benchmark study. The DenseNet-201 and InceptionResNetV2 models, employing deep transfer learning to optimize training on limited data, detected abnormalities in the humerus radiographs with 95% CI accuracies of 83–92% and high sensitivities greater than 0.9, allowing for these models to serve as useful initial screening tools to prioritize studies for expedited review. The performance in the case of finger radiographs was not as promising, possibly due to the limitations of large inter-radiologist variation. It is suggested that the causes of this variation be further explored using machine learning approaches, which may lead to appropriate remediation.

scholarly journals Optimising Seniors’ Metabolism of Medications and Avoiding Adverse Drug Events Using Data on How Metabolism by Their P450 Enzymes Varies with Ancestry and Drug–Drug and Drug–Drug–Gene Interactions

2020 ◽  
Vol 10 (3) ◽  
pp. 84
Author(s):  
Roger E. Thomas
Keyword(s):  
Primary Care ◽  
Primary Care Physicians ◽  
Adverse Drug Events ◽  
Meta Analysis ◽  
Gene Interactions ◽  
Rural And Remote ◽  
Remote Areas ◽  
Rural And Remote Areas ◽  
Enormous Variation ◽  
Using Data

Many individuals ≥65 have multiple illnesses and polypharmacy. Primary care physicians prescribe >70% of their medications and renew specialists’ prescriptions. Seventy-five percent of all medications are metabolised by P450 cytochrome enzymes. This article provides unique detailed tables how to avoid adverse drug events and optimise prescribing based on two key databases. DrugBank is a detailed database of 13,000 medications and both the P450 and other complex pathways that metabolise them. The Flockhart Tables are detailed lists of the P450 enzymes and also include all the medications which inhibit or induce metabolism by P450 cytochrome enzymes, which can result in undertreatment, overtreatment, or potentially toxic levels. Humans have used medications for a few decades and these enzymes have not been subject to evolutionary pressure. Thus, there is enormous variation in enzymatic functioning and by ancestry. Differences for ancestry groups in genetic metabolism based on a worldwide meta-analysis are discussed and this article provides advice how to prescribe for individuals of different ancestry. Prescribing advice from two key organisations, the Dutch Pharmacogenetics Working Group and the Clinical Pharmacogenetics Implementation Consortium is summarised. Currently, detailed pharmacogenomic advice is only available in some specialist clinics in major hospitals. However, this article provides detailed pharmacogenomic advice for primary care and other physicians and also physicians working in rural and remote areas worldwide. Physicians could quickly search the tables for the medications they intend to prescribe.

scholarly journals Sources of Information on Medicinal Products Among Physicians – A Survey Conducted Among Primary Care Physicians in Poland

2022 ◽  
Vol 12 ◽  
Author(s):  
Magdalena Zielińska ◽  
Tomasz Hermanowski
Keyword(s):  
Primary Care ◽  
Clinical Guidelines ◽  
Primary Care Physicians ◽  
Demographic Data ◽  
Sources Of Information ◽  
Medicinal Products ◽  
Using Data ◽  
Credible Information ◽  
Source Of Information ◽  
Credible Source

Introduction: Primary care physicians need to have access to up-to-date knowledge in various fields of medicine and high-quality information sources, but little is known about the use and credibility of sources of information on medicinal products among Polish doctors. The main goal of this study was to analyze the sources of information on medicinal products among primary care physicians in Poland.Methods: A survey was conducted among 316 primary care physicians in Poland. The following information was collected: demographic data of participants, type and frequency of using data sources on medicinal products, barriers to access credible information, assessment of the credibility of the sources used, impact of a given source and other factors on prescription decisions.Results: The most frequently mentioned sources of information were medical representatives (79%), medical journals (78%) and congresses, conventions, conferences, and training (76%). The greatest difficulty in finding the latest information about medicinal products was the lack of time. The surveyed doctors considered clinical guidelines to be the most credible source of information, and this source also had the greatest impact on the choice of prescribed medicinal products.Conclusion: The study showed that clinicians consider clinical guidelines as the most credible source of information with the greatest impact on prescribing medicinal products. However, it is not the source most often mentioned by doctors for obtaining knowledge about medicinal products. There is a need to develop strategies and tools to provide physicians with credible sources of information.

The Pharmacist Practitioner Training Program

1980 ◽  
Vol 14 (2) ◽  
pp. 114-119 ◽  
Author(s):  
Gary P. Copeland ◽  
David A. Apgar
Keyword(s):  
Primary Care ◽  
Health Service ◽  
Physical Examination ◽  
Training Program ◽  
Clinical Setting ◽  
Chronic Illnesses ◽  
Indian Health Service ◽  
Indian Health ◽  
Practitioner Training ◽  
Medical Histories

Since 1973, the Indian Health Service has been training pharmacists to provide primary care to patients in an expanded role, including the compilation of complete medical histories, physical examination, diagnosis, and treatment of outpatients with selected acute and chronic illnesses. This article discusses the evolution of the Pharmacist Practitioner Training Program, the training and experience received through this program, and the utilization of graduates in the clinical setting.

scholarly journals Moving the needle towards the democratization of echocardiography: a case report

2019 ◽  
Author(s):  
Jonathan dos Santos ◽  
Patrícia Borges Fernandes ◽  
Francisco Rocha Gonçalves ◽  
Alexandra Gonçalves
Keyword(s):  
Primary Care ◽  
Physical Examination ◽  
Primary Care Physician ◽  
Primary Care Physicians ◽  
Point Of Care ◽  
Family Doctor ◽  
Vena Cava ◽  
Left Ventricular ◽  
Care Physician ◽  
Point Of Care Ultrasound

Abstract Background Echocardiography has been traditionally performed in echo labs and the potential benefits of its use by primary care physicians (PCPs) are still unexplored. We present a case where POCUS (point-of-care ultrasound) was used as a complement of physical examination by a family doctor, allowing a prompt clinical decision in a heart failure (HF) patient. Case summary An 85-year-old woman, living independently, asks her family doctor for a home consultation due to increasing dyspnoea. On examination, severe dyspnoea and bilateral ankle oedema was noted and a point-of-care echocardiogram was performed by the primary care physician, who observed: severely compromised left ventricular systolic function, moderate mitral and tricuspid regurgitation, and severe dilation of the inferior vena cava. As a result, the diagnosis of HF with decreased ejection fraction was formed supporting the therapeutic decision. Discussion This case represents an elderly patient with dyspnoea, without previous HF diagnosis. The primary care physician, used portable ultrasound as a complement of physical examination, which confirmed a HF diagnosis, allowing a prompt decision-making on therapy. POCUS, can be a powerful tool to expedite treatment in different settings, including the home consultations by PCPs.

Monitoring the Physical Health of Cancer Survivors: A Survivorship-Focused Medical History

2006 ◽  
Vol 24 (32) ◽  
pp. 5105-5111 ◽  
Author(s):  
Patricia A. Ganz
Keyword(s):  
Primary Care ◽  
Health Care ◽  
Cancer Survivors ◽  
Medical History ◽  
Health Care Providers ◽  
Primary Care Physicians ◽  
Care Providers ◽  
Follow Up Care ◽  
Physical Effects

Cancer survivors frequently visit their primary-care physicians, as well as oncology specialists, for follow-up care. There is a need to monitor these survivors for the late physical effects of cancer, yet few health care providers have received training in how to do this. This article provides guidance on how to take a cancer survivor-directed medical history to facilitate the elicitation of relevant exposures, family history, and symptoms that may be related to the late effects of cancer therapy.

Confirmation of Dermatophytes in Nail Specimens Using In-Office Dermatophyte Test Medium Cultures

2003 ◽  
Vol 93 (3) ◽  
pp. 195-202 ◽  
Author(s):  
Maureen B. Jennings ◽  
Michael G. Rinaldi
Keyword(s):  
Primary Care ◽  
Convenient Method ◽  
Primary Care Physicians ◽  
Laboratory Tests ◽  
Test Results ◽  
Central Laboratory ◽  
Test Medium ◽  
Laboratory Cultures ◽  
Special Expertise ◽  
Using Data

Using data from a multicenter nationwide multispecialty survey, the authors investigated the efficacy of in-office dermatophyte test medium (DTM) and central laboratory cultures used to confirm onychomycosis across samples collected by podiatric, dermatologic, and primary-care physicians. The samples collected by podiatric physicians were both positive or both negative in 43% and 27% of patients, respectively. Samples harvested by dermatologists were both positive in 37% of patients and both negative in 32%, while the samples collected by primary-care physicians were both positive in 28% of patients and both negative in 38%. The accuracy of DTM and central laboratory tests is dependent on the proper collection of nail samples, and the accuracy of mycologic test results varied significantly across nail specimens harvested by podiatric, dermatologic, and primary-care physicians. DTM culture was found to be an effective and convenient method of confirming dermatophyte infections in patients with signs of onychomycosis. The data presented here indicate that the special expertise of podiatric physicians in treating foot-related illnesses translates into more accurate mycologic testing. (J Am Podiatr Med Assoc 93(3): 195-202, 2003)

Sign in / Sign up

Export Citation Format

Share Document