Den här webbplatsen använder kakor

För att webbplatsen ska fungera på ett bra sätt för dig använder vi nödvändiga kakor (cookies). Vi använder också kakor för webbanalys för att förbättra webbplatsen.

Läs mer om hur vi använder cookies

Meny

AI-Created Diagnostic Scores for Emergency Department Patients with Breathing Difficulties

Ellen Tolestam Heyman

Content

Background

Half of all adult emergency department (ED) visits with a complaint of breathing difficulties, or dyspnoea, involve acute heart failure (AHF), exacerbation of chronic obstructive pulmonary disease (eCOPD), or pneumonia, which are often misdiagnosed and mistreated. We aimed to create an artificial intelligence (AI) diagnostic decision support tool to detect patients with AHF, eCOPD, and pneumonia among dyspnoeic adults at the beginning of their ED visit, before any blood tests, imaging, and physician assessment.

Methods

In this cross-sectional study, we included all ED visits of patients aged 18 years or older with dyspnoea at two regional Swedish EDs between 01/07/2017 and 31/12/2019. Gold standard diagnoses were defined as the subsequent in-hospital or ED discharge notes, and a subsample was manually reviewed by emergency medicine experts. We developed a novel neural network model and analysed unselected data from a complete regional healthcare system, along with ECGs and socioeconomic factors, by taking the event time and clinical context of the data into account. Cohort performance was compared to a simpler CatBoost model.

Results

We included 10,869 visits, with 15.1% having AHF, 13.6% eCOPD, and 13.1% pneumonia. Among the 11,656 variables, the median number of unique variables selected per ED visit was 187 (interquartile range 111–307). The model’s performance had a median micro AUROC of 87.8% (2.5–97.5 percentile; 86.4–89.3%), compared to the simpler CatBoost baseline model AUROC of 81.4% (77.5–86.6%). Aggregating the unique sets of variables
resulted in a cohort list of 2,064 diagnostic variables. Age, ECGs, previous diagnoses, and medication were considered important by the AI model, while sex, vital signs, and socioeconomic factors were deemed almost non-predictive.

Interpretation

By analysing all data, more aspects of data, and individualised variable sets through neural networks, we personalise and improve diagnostics in the emergency department. Our model
is generic and might be tested for other complaints and diagnoses.

Further information

At the bottom, the top 100 diagnostic variables of the AI generated variable list are shown.

Researchers

Ellen Tolestam Heyman (1,2), Awais Ashfaq (3,4), Ulf Ekelund (2,5), Mattias Ohlsson (4,6), Jonas Björk (7,8), Alexander Marcel Schubert (9), Ardavan M. Khoshnood (10,11), Markus Lingman (3,4,12)

(1) Department of Emergency Medicine, Halland Hospital, Region Halland, Sweden

(2) Emergency Medicine, Department of Clinical Sciences Lund, Lund University, Lund, Sweden

(3) Halland Hospital, Region Halland, Sweden

(4) Center for Applied Intelligent Systems Research (CAISR), Halmstad University, Halmstad, Sweden

(5) Skåne University Hospital, Lund, Sweden

(6) Centre for Environmental and Climate Science, Lund University, Lund, Sweden

(7) Department of Laboratory Medicine, Division of Occupational and Environmental Medicine, Lund University, Lund, Sweden

(8) Clinical Studies Sweden, Forum South, Skåne University Hospital, Lund, Sweden

(9) Department of Computational Precision Health, University of California, Berkeley, USA, and Department of Computational Precision Health, University of California, San Francisco, USA

(10) Emergency Medicine, Department of Clinical Sciences Malmö, Lund University, Lund, Sweden

(11) Skåne University Hospital, Malmö, Sweden

(12) Department of Molecular and Clinical Medicine, Institute of Medicine, Sahlgrenska Academy, University of Gothenburg, Gothenburg, Sweden

Funding

These studies were part of the AIR Lund (Artificially Intelligent use of Registers at Lund University) research network in Lund, Sweden. The work was funded by the Swedish Research Council; the Scientific Council of Region Halland, Sweden; Sparbanksstiftelsen Varberg, Sweden; and the foundation Stiftelsen Landshövding Per Westlings Minnesfond, Sweden. The funders had no role in the study design, data collection, analysis, interpretation, writing of the report, or decision to publish.

Partners

Region Halland, Halmstad University, Lund University

Heyman Tolestam, E., Ashfaq, A., Ekelund U., Ohlsson, M., Björk, J., Khoshnood, A., & Lingman, M. A novel interpretable deep learning model for diagnosis in emergency department dyspnoea patients based on complete data from an entire health care system. Accepted in PLOS One.

Heyman Tolestam, E., Ashfaq, A., Ekelund U., Ohlsson, M., Björk, J., Schubert, A. M., Lingman, M., & Khoshnood, A. Utilizing artificial intelligence and medical experts to identify predictors for common diagnoses in dyspnoeic adults: A cross-sectional study of consecutive emergency department patients from southern Sweden. Manuscript under review.

The top 100 variables were selected from an AI-generated list of 2,064 diagnostic variables, drawn from a total of 11,656 variables. The remaining variables were considered noise. The list was created by aggregating individual patient variable sets.

  1. Chronic obstructive pulmonary disease (COPD), primary diagnosis
  2. Heart failure, primary diagnosis
  3. ECG: Atrial fibrillation (unknown atrial activity)
  4. Chronic obstructive pulmonary disease (COPD), secondary diagnosis
  5. Age > 95 years
  6. Age 90-94 years
  7. ECG: Ventricular-paced complexes, other complexes also detected
  8. ECG: Left Ventricular Hypertrophy with secondary repolarization abnormality (multi-LVH criteria, repolarisation abnormality)
  9. Chronic obstructive pulmonary disease (COPD), primary care complaint
  10. ECG: Atrial fibrillation (ventricular rate **-**, irregular ventricular activity)
  11. Atrioventricular and left bundle-branch block, secondary diagnosis
  12. Age 55-59 years
  13. Heart failure, secondary diagnosis
  14. Veterinary medication for cardiovascular system, collected medication
  15. Cutaneous abscess, furuncle and carbuncle, secondary diagnosis
  16. Age 85-89 years
  17. Diseases of vocal cords and larynx, not elsewhere classified, secondary diagnosis
  18. Medication against obstructive airways, collected medication
  19. Age 70-74 years
  20. ECG: Anterior Q waves, possibly due to LVH (Q >30mS, V1 V2 & LVH)
  21. Age 75-79 years
  22. Age 80-84 years
  23. "Heart failure", primary care and outpatient specialist care complaint
  24. "Amputation", primary care and outpatient specialist care complaint
  25. Age 45-49 years
  26. ECG: Abnormal T, consider ischemia, diffuse leads (T <-0.20mV, ant/lat/inf)
  27. Hypertensive chronic kidney disease, secondary diagnosis
  28. Atrial fibrillation and flutter, secondary diagnosis
  29. Other gynecological, collected medication
  30. Non-pressure chronic ulcer of lower limb, not elsewhere classified, primary diagnosis
  31. Age 65-69 years
  32. ECG: Repolarisation abnormality, severe global ischemia ((LM/3VD) STe aVR, STd & Tneg, ant/lat/in)
  33. Age 50-54 years
  34. ECG: Atrial-sensed ventricular-paced complexes (other complexes also detected)
  35. "Makula", primary care and outpatient specialist care complaint
  36. Antihypertensives, prescribed medication
  37. Retention of urine, secondary diagnosis
  38. "Aid for personal needs", primary care and outpatient specialist care complaint
  39. Influenza due to identified seasonal influenza virus, secondary diagnosis
  40. Fracture of lower leg, including ankle, primary diagnosis
  41. ECG: Repolarisation abnormality suggests ischemia, diffuse leads (ST-T neg, ant/lat/inf)
  42. Type 2 diabetes mellitus, primary diagnosis
  43. "Assessment", primary care and outpatient specialist care complaint
  44. ECG: Abnormal T, consider ischemia, lateral leads (T <-0.20mV, I aVL V5 V6)
  45. Other peripheral vascular diseases, secondary diagnosis
  46. ECG: Intraventricular conduction delay, consider atypical LBBB (QRSd> ** , notch/slur R I aVL V5-6)
  47. ECG: Sinus or ectopic atrial rhythm (P axis (-45,135))
  48. ECG: Left bundle-branch block (QRSd> ** , broad/notched R)
  49. Lab: Digoxin within normal range
  50. Age 60-64 years
  51. Antigout preparations, collected medication
  52. Mental and behavioral disorders due to use of opioids, primary diagnosis
  53. ECG: Atrial-paced complexes (other complexes also detected)
  54. ECG: Repolarisation abnormality suggests ischemia, lateral leads (ST dep, T neg, I aVL V5 V6)
  55. Fracture of forearm, primary diagnosis
  56. Diabetes, chiropody, primary care and outpatient specialist care complaint*
  57. Neoplasm of uncertain behavior of urinary organs, primary diagnosis
  58. Gastric ulcer, primary diagnosis
  59. Other anemia, secondary diagnosis
  60. "Patient with ICD" (implantable cardioverter-defibrillator), primary care and outpatient specialist care complaint
  61. Antigout preparations, prescribed medication
  62. ECG: Paired ventricular premature complexes (sequence of 2 V complexes)
  63. "Prothrombin time" (PT), primary care and outpatient specialist care complaint
  64. "Chiropody", primary care and outpatient specialist care complaint
  65. ECG: Anterior ST elevation, probably due to LVH (ST >0.20 mV in V1-V4 & LVH)
  66. Rheumatic tricuspid valve diseases, secondary diagnosis
  67. Other gynecological, prescribed medication
  68. Complications of internal orthopedic prosthetic devices, implants and grafts, secondary diagnosis
  69. "Care planning", primary care and outpatient specialist care complaint
  70. Other and unspecified dorsopathies, not elsewhere classified, secondary diagnosis
  71. Osteoarthritis of hip, primary diagnosis
  72. Nonrheumatic mitral valve disorders, secondary diagnosis
  73. "Nose bleeding", ED complaint
  74. ECG: Nonspecific T abnormalities, lateral leads (T <-0.10mV, I aVL V5 V6)
  75. Erysipelas, primary diagnosis
  76. Cytology, primary care and outpatient specialist care complaint*
  77. ECG: Ventricular bigeminy (bigeminy string>4 w/ V complexes)
  78. ECG: Nonspecific repolarisation abnormality, diffuse leads (ST dep, T flat/neg, ant/lat/inf)
  79. Problems related to lifestyle, secondary diagnosis
  80. Lab: Creatinine above normal range
  81. Measurement of blood pressure in toe or finger
  82. Other diseases of the digestive system, primary diagnosis
  83. Dislocation and sprain of joints and ligaments at ankle, foot and toe level, primary diagnosis
  84. “Check-up", primary care and outpatient specialist care complaint
  85. ECG: No further rhythm analysis attempted due to paced rhythm
  86. Symptoms and signs concerning food and fluid intake, secondary diagnosis
  87. Atopic dermatitis, primary diagnosis
  88. Right side
  89. ECG: Minimal ST depression, lateral leads (ST <-0.04mV, I aVL V5 V6)
  90. Radiology: X-ray of elbow
  91. Emphysema, secondary diagnosis
  92. Paralytic ileus and intestinal obstruction without hernia, primary diagnosis
  93. ECG: Repolarization abnormality, probably rate related (ST depression, T-negativity, tachycardia)
  94. ECG: Anterior infarct, old (Q >30mS, abnormal ST-T, V2-V5)
  95. "Wound", primary care and outpatient specialist care complaint
  96. Lab: Prothrombin time (PT) above normal range
  97. Bacterial pneumonia, not elsewhere classified, secondary diagnosis
  98. ECG: Ventricular premature complex (V complex w/ short R-R interval)
  99. Triage: Oxygen saturation 80-85% (regardless of oxygen gas treatment)
  100. Unspecified hematuria, secondary diagnosis

Senast ändrad: