covid.datahelps.life

Research

We are keen to collaborate on the following aspects. Please get in touch if you are interested: honghan.wu@ucl.ac.uk.

  1. externally validate our models in different settings. Model technical details listed below.
  2. validate other research groups’ models in our datasets.
  3. jointly create a federated COVID-19 dataset (via a bring-model-to-data pipeline) for supporting community efforts on tackling COVID-19.

Research resources

Below are technical details (coefficients and intercepts of logistic regression models) of our prediction models for Death and Poor Outcome.

Prediction models for ‘death’

Data in the table head are number (%). Number is the number patients with complete data of the corresponding risk group. % is the percentage of patients with the outcome. Data in the table cells are coeffcient (95%CI). ·· indicates the corresponding predictor is not used. Initials for specifying predictor groups D - Demographic C - Underlying Conditions S - Symptoms L - Laboratory Results


Predictor

DCS 30/769 (3.9%)
Death
DCSL 19/651 (2.9%)

DL 20/653 (3.1%)
Demographic      
Age 0.0461 (0.0122:0.0799) -0.0020 (-0.0531:0.0491) 0.0142 (-0.0370:0.0654)
Sex (male) 0.3177 (-0.4743:1.1097) -0.8563 (-2.1144:0.4019) -0.8035 (-2.0102:0.4031)
Premorbid Conditions      
Chronic lung disease 1.0975 (0.0706:2.1243) 0.4054 (-1.3315:2.1424) ··
Diabetes Mellitus 0.4448 (-0.5191:1.4087) 0.3168 (-1.0493:1.6830) ··
Immunocompromised 1.3798 (-0.4451:3.2047) ·· ··
Malignancy 0.0185 (-1.9155:1.9526) -0.0019 (-2.0489:2.0452) ··
Hypertension 0.3457 (-0.4809:1.1722) ·· ··
Heart disease 0.6761 (-0.2613:1.6135) ·· ··
Chronic renal disease 1.3210 (-0.0856:2.7276) ·· ··
Symptoms      
Fever ·· ·· ··
Cough 0.7409 (-0.3614:1.8431) 0.9446 (-0.6341:2.5233) ··
Fatigue ·· ·· ··
Dyspnoea 1.5245 (0.5816:2.4675) 0.8559 (-0.4246:2.1365) ··
Diarrhoea 0.0000 (-2.1908:2.1908) ·· ··
Admission clinical features      
Neutrophil count (10^9/L) ·· 0.1793 (0.0294:0.3292) 0.2249 (0.0844:0.3654)
Lymphocyte count (10^9/L) ·· -1.4449 (-2.9297:0.0400) -1.3255 (-2.7882:0.1372)
Platelet count (10^9/L) ·· -0.0059 (-0.0129:0.0011) -0.0052 (-0.0118:0.0015)
C-reactive protein (mg/L) ·· 0.0140 (0.0028:0.0251) 0.0138 (0.0035:0.0240)
Creatinine (μmol/L) ·· 0.0012 (-0.0023:0.0047) 0.0015 (-0.0020:0.0049)
Intercept -8.5016 -3.2048 -3.5172

Prediction models for ‘poor outcome’


Predictor

DCS 72/768 (9.4%)
Poor outcome
DCSL 57/651 (8.8%)

DL 58/653 (8.9%)
 
Demographic        
Age 0.0414 (0.0187:0.0640) 0.0182 (-0.0145:0.0509) 0.0281 (-0.0004:0.0565)  
Sex (male) 0.2050 (-0.3270:0.7370) -0.5136 (-1.2856:0.2583) -0.4723 (-1.1882:0.2436)  
Premorbid Conditions        
Chronic lung disease 1.2284 (0.4757:1.9812) 0.9657 (-0.2093:2.1408) ··    
Diabetes Mellitus 0.0000 (-0.6983:0.6983) -0.2920 (-1.2810:0.6971) ··  
Immunocompromised 1.4338 (0.0494:2.8182) 1.5738 (-0.4570:3.6045) ··  
Malignancy ·· 0.0000 (-1.3836:1.3836) -0.7171 (-2.4933:1.0591) ··
Hypertension 0.6631 (0.1047:1.2215) 0.4819 (-0.3166:1.2804) ··  
Heart disease 0.7408 (0.0745:1.4070) 0.7360 (-0.2202:1.6922) ··  
Chronic renal disease 0.9920 (-0.1025:2.0865) -0.3503 (-2.2309:1.5304) ··  
Symptoms        
Fever 0.0280 (-0.5615:0.6175) 0.1678 (-0.6413:0.9769) ··  
Cough -0.0000 (-0.6305:0.6305) -0.6005 (-1.4021:0.2012) ··  
Fatigue 0.0996 (-0.4731:0.6724) 0.2024 (-0.5748:0.9796) ··  
Dyspnoea 1.0902 (0.4967:1.6837) 1.2438 (0.4404:2.0472) ··  
Diarrhoea -0.2074 (-1.7452:1.3304) -2.6571 (-8.2778:2.9636) ··  
Admission clinical features        
Neutrophil count (10^9/L) ·· 0.2489 (0.1114:0.3864) 0.2575 (0.1340:0.3810)  
Lymphocyte count (10^9/L) ·· -1.2256 (-2.1206:-0.3306) -1.1842 (-1.9857:-0.3827)  
Platelet count (10^9/L) ·· -0.0036 (-0.0081:0.0009) -0.0027 (-0.0068:0.0013)  
C-reactive protein (mg/L) ·· 0.0078 (-0.0020:0.0176) 0.0082 (-0.0003:0.0167)  
Creatinine (μmol/L) ·· 0.0125 (0.0045:0.0204) 0.0099 (0.0036:0.0161)  
Intercept -6.2993 -4.2863 -4.1865