System and method for quantitative assessment of frailty
Methods, systems, and apparatus for quantifying an individual's frailty level based on inertial sensor data collected from the individual. The quantified frailty level may correspond to and approximate clinical metrics of frailty, such as the Fried frailty index. A linear regression model may be used to output the quantitative frailty value based on input parameters from the inertial sensor data. The linear regression model may be initially generated from the clinically-measured frailty index values of individuals and inertial sensor data collected from them. The inertial sensor data may be collected during, for example, a timed up and go (TUG) test. Two logistic regression models may be used to output a frailty class based on input parameters from the inertial sensor data. A first logistic regression model may distinguish between robust and frail individuals. A second logistic regression model may distinguish between robust and pre-frail individuals.
1. A computer-implemented method for estimating frailty, the method comprising:
receiving a first reference frailty index value associated with a first individual, a second reference frailty index value associated with a second individual;
receiving a first set of inertial sensor data associated with the first individual and a second set of inertial sensor data associated with the second individual, wherein the first set of inertial sensor data and second set of inertial sensor data comprise angular velocity data and acceleration data;
generating a linear regression model that outputs a first frailty value estimate based on the first set of inertial sensor data and that outputs a second frailty value estimate based on the second set of inertial sensor data; and
storing the linear regression model in a non-transitory computer-readable medium,
wherein the linear regression model is based on the difference between the first frailty value estimate and the first reference index value and based on the difference between the second frailty value estimate and the second reference index value.
2. The method of claim 1 , further comprising:
receiving a third reference index value associated with a third individual; and
receiving a third set of inertial sensor data, wherein the linear regression model is configured to output a third frailty value estimate based on the third set of inertial sensor data, and wherein the linear regression model is further based on the difference between the third frailty value estimate and the third reference index value.
3. The method of claim 2 , further comprising:
associating the first reference frailty index value with a reference robust class and associating the second reference frailty index value with a reference frail class, wherein the first reference frailty index value is different than the second reference frailty index value; and
generating a first logistic regression model that outputs a first frailty class estimate based on the first set of inertial sensor data and that outputs a second frailty class estimate based on the second set of inertial sensor data,
wherein the first logistic regression model is based on one or more differences between the first set of inertial sensor data and the second set of inertial sensor data.
4. The method of claim 3 , further comprising:
associating the third reference index value with a pre-frail class, wherein the third reference index value is different than the first reference index value and the second reference index value; and
generating a second logistic regression model that outputs a third frailty class estimate based on the third set of inertial sensor data,
wherein the second logistic regression model is based on one or more differences between the first set of inertial sensor data and the third set of inertial sensor data.
5. An apparatus comprising one or more processors, the one or more processors configured to
receive a first reference frailty index value associated with a first individual, a second reference frailty index value associated with a second individual;
receive a first set of inertial sensor data associated with the first individual and a second set of inertial sensor data associated with the second individual, wherein the first set of inertial sensor data and second set of inertial sensor data comprise angular velocity data and acceleration data;
generate a linear regression model that outputs a first frailty value estimate based on the first set of inertial sensor data and that outputs a second frailty value estimate based on the second set of inertial sensor data; and
store the linear regression model in a non-transitory computer-readable medium,
wherein the linear regression model is based on the difference between the first frailty value estimate and the first reference index value and based on the difference between the second frailty value estimate and the second reference index value.
6. The apparatus of claim 5 , wherein the one or more processors are configured to:
receive a third reference index value associated with a third individual; and
receive a third set of inertial sensor data, wherein the linear regression model is configured to output a third frailty value estimate based on the third set of inertial sensor data, and wherein the linear regression model is further based on the difference between the third frailty value estimate and the third reference index value.
7. The apparatus of claim 6 , wherein the processors are further configured to:
associate the first reference frailty index value with a reference robust class and associate the second reference frailty index value with a reference frail class, wherein the first reference frailty index value is different than the second reference frailty index value; and
generate a first logistic regression model that outputs a first frailty class estimate based on the first set of inertial sensor data and that outputs a second frailty class estimate based on the second set of inertial sensor data,
wherein the first logistic regression model is based on one or more differences between the first set of inertial sensor data and the second set of inertial sensor data.
8. The apparatus of claim 7 , wherein the one or more processors are further configured to:
associate the third reference index value with a pre-frail class, wherein the third reference index value is different than the first reference index value and the second reference index value; and
generate a second logistic regression model that outputs a third frailty class estimate based on the third set of inertial sensor data,
wherein the second logistic regression model is based on one or more differences between the first set of inertial sensor data and the third set of inertial sensor data.