Biomarkers – Predictive Models for COVID-19

Use of Self-reported Symptoms via Smartphone App to Predict Potential COVID-19 Infection

The aim of this study was to identify the combination of symptoms most predictive of COVID-19.

This should help guiding recommendations for self-isolation and prevent further spread of the disease. Unfortunately, COVID-19 testing is not accessible to the majority of the population.

Participants and Methods

2,618,862 participants (UK and US) reported their symptoms on a smartphone-based app.

18,401 had undergone a SARS-CoV-2 test. Self-reported symptoms via a free smartphone app (launched in the UK and US on 24 and 29 March 2020). Individuals reported health information on a daily basis, including symptoms, hospitalization, PCR (RT-PCR) test outcomes, demographic information and pre-existing medical conditions.

Results (COVID-19 Tested Individuals)

Based on the UK data from tested individuals, loss of smell was associated to odds ratio (OR)=6.40; 95% confidence interval (CI)=5.96–6.87; P<0.0001 (see table).

Similarly in US data loss of smell was associated to OR=10.01; 95% CI=8.23–12.16; P<0.0001 (see table). Results are after adjusting for age, sex and body mass index. Combined datasets (UK and US) and the adjusted results using inverse variance fixed-effects meta-analysis gave an OR=6.74; 95% CI=6.31–7.21; P<0.0001.

Prediction Model

A combination of loss of smell, fatigue, persistent cough and loss of appetite resulted in the best model. Stratification for sex and age groups found no differences.

UK cohort: Sensitivity of 0.65 (0.62–0.67) and specificity of 0.78 (0.76–0.80).

US cohort: Sensitivity of 0.66 (0.62–0.69), a specificity of 0.83 (0.82–0.85).

The predictive model was applied to 805,753 UK and US symptom-reporting individuals (not tested for COVID-19): 17.42% (14.45– 20.39%) of individuals reporting symptoms were likely to be infected by the virus, representing 5.36% of overall responders to the app.

Limitations of the Study

Based on self reported symptoms and restricted to app users.

No generalized testing but focused on symptomatic individuals.

More about the app: COVID Symptom Tracker mobile app was created in collaboration with Zoe Global Ltd, a digital healthcare company, and academic scientists from Massachusetts General Hospital and King’s College London. (

Hereditability of Self-reported COVID-19 Symptoms
- Results Coming Soon

TwinsUK volunteers (n=2,633) completing the C-19 Covid symptom tracker app (Mz and Dz twins).

Heritability for self-reported symptoms, using models of additive genetic (A), shared environment (C), unique environment (E).

To be followed up after peer-review; posted April 27 (Tim Spector group).