Ambulatory system for measuring and monitoring physical activity and risk of falling and for automatic fall detection

A method for monitoring body movement, the method comprising:

• receiving acceleration measurements corresponding to one or more acceleration signals generated by at least one accelerometer component of a sensor unit attached to an upper part of a body of a person in response to movement of the body;
• with a data processing system comprising one or more processor circuits:
• applying a filter to at least a portion of at least one of the one or more acceleration signals to obtain one or more filtered signals;
• determining an estimated postural transition duration for a postural transition based at least partly on one or more peaks in the one or more filtered signals;
• deriving information related to the postural transition using the estimated postural transition duration, the information including at least one of:
• time of occurrence of the postural transition, duration of the postural transition, or type of the postural transition;
• determining a revised estimated postural transition duration that is more accurate than the estimated postural transition duration;
• and
• deriving at least part of the information by estimating an acceleration of the person during a time period corresponding to the revised estimated postural transition duration, said estimating comprising:
• processing a frontal acceleration signal of the one or more acceleration signals to correct for a sagittal trunk tilt angle of the person during the time period corresponding to the revised estimated postural transition duration to obtain a corrected frontal acceleration signal;
• using the corrected frontal acceleration signal and a vertical acceleration signal of the one or more acceleration signals, estimating time-varying inertial frontal and vertical acceleration signals;
• removing a gravitational component from the estimated time-varying inertial vertical acceleration signal;
• computing a time-varying magnitude signal corresponding to a square root of a sum of squares of said estimated time-varying inertial frontal and vertical acceleration signals;
• applying at least one filter to said time-varying magnitude signal to obtain a filtered time-varying magnitude signal;
• identifying a local extremum in said filtered time-varying magnitude signal preceding the time of occurrence of said postural transition (P_pre);
• and
• identifying a local extremum in said filtered time-varying magnitude signal following the time of occurrence of said postural transition (P_post).

Non-transitory computer storage that stores executable code that directs a computer system to at least:

• apply a filter to at least a portion of at least one of one or more acceleration signals generated by at least one accelerometer component of a sensor unit attached to an upper part of a body of a person, the one or more acceleration signals generated in response to movement of the body;
• determine an estimated postural transition duration for a postural transition based at least partly on one or more peaks in the at least a portion of the one or more acceleration signals that is filtered;
• using the estimated postural transition duration, derive information related to the postural transition, the information including at least one of:
• time of occurrence of the postural transition, duration of the postural transition, or type of the postural transition;
• apply a filter to a vertical accelerometer signal of the one or more acceleration signals for a time period corresponding to said duration of said postural transition, to obtain a filtered vertical acceleration signal;
• compute a gradient of the filtered vertical acceleration signal to obtain a gradient signal;
• identify a global extremum of said gradient signal;
• if the magnitude of the global extremum exceeds a threshold value, compare a first portion of the filtered vertical acceleration signal that precedes the global extremum to a second portion of the filtered vertical acceleration signal that is subsequent to the global extremum;
• if said comparison yields a first result, identify a lying posture;
• and
• if said comparison yields a second result, identify a transition from said lying posture.