“Phenotypic Ages” in NHANES Data

This example loads blood exam data from NHANES 2010, calculates “Phenotypic Ages,” and performs survival analyses by phenotypic age.

Loading NHANES 2010 data

from biolearn.load import load_nhanes
year = 2010
df = load_nhanes(year)
df["years_until_death"] = df["months_until_death"] / 12

Calculate “biological age” based on PhenotypicAge

from biolearn.hematology import phenotypic_age
df["phenotypic_age"]=phenotypic_age(df)

Show relation between biological age and chronological age

import seaborn as sn
sn.scatterplot(data=df,x="age", y="phenotypic_age",s=2);

<Axes: xlabel='age', ylabel='phenotypic_age'>

Plot survival curve for people with accelerated aging (older biological age) vs decelerated aging (younger biological age)

import matplotlib.pyplot as plt
from lifelines import KaplanMeierFitter
df["biologically_older"] = df["phenotypic_age"] > df["age"]
ax = plt.subplot()
groups = df["biologically_older"]
ix = groups == 0
T = df.years_until_death
E = df.is_dead
kmf = KaplanMeierFitter()
kmf.fit(T[ix], E[ix], label="Biologically younger")
ax = kmf.plot_survival_function(ax=ax)
kmf.fit(T[~ix], E[~ix], label="Biologically older")
ax = kmf.plot_survival_function()
plt.ylabel("Survival")
plt.xlabel("Years");

Text(0.5, 23.52222222222222, 'Years')

Estimated memory usage: 129 MB