Physiological property forecasting
Inventors
HAYERI, Amir Hossein • CACHO, Ricardo • RAMCHANDANI, Suvan Ashok • BRANDT, Nicolas • SHER, Aleksey
Assignees
Interested in licensing this patent?
MTEC can help explore whether this patent might be available for licensing for your application.
Publication Number
US-12220263-B2
Publication Date
2025-02-11
Expiration Date
Abstract
A computer-implemented method of facilitating physiological property forecasting for detecting disease complications is disclosed. The method involves receiving signals representing sensed physiological property indicators, each of the sensed physiological property indicators representing a sensed physiological property of a patient at a respective time and receiving signals representing one or more contextual indicators associated with the sensed physiological property indicators. The method also involves applying at least one classification criterion to the one or more contextual indicators to determine a patient state of a plurality of possible patient states, each patient state associated with a respective set of forecasting parameters and applying the set of forecasting parameters associated with the determined patient state to the sensed physiological property indicators to determine at least one forecast physiological property indicator representing a forecast physiological property of the patient at a future time. Other methods, systems and computer-readable media are also disclosed.
Core Innovation
The invention relates to a computer-implemented physiological-property forecasting framework for detecting disease complications. The method receives signals representing sensed physiological property indicators of a patient at respective times and signals representing one or more contextual indicators associated with the sensed physiological property indicators. At least one classification criterion is applied to the contextual indicators to determine a patient state of a plurality of possible patient states, where each patient state is associated with a respective set of forecasting parameters.
For the determined patient state, the method applies the associated set of forecasting parameters to the sensed physiological property indicators to determine at least one forecast physiological property indicator representing a forecast physiological property at a future time. Each set of forecasting parameters includes a plurality of sets of historical physiological property indicators representing physiological properties of the patient during respective historical time periods. The method compares the sensed physiological property indicators to the historical physiological property indicators, selects at least one historical set based on the comparing, generates a time-dependent function representing the selected historical set, and determines the forecast using the generated time-dependent function.
The comparing includes determining a sum of differences between the sensed physiological property indicators and respective ones of the historical physiological property indicators, and optionally determining a weight associated with the historical physiological property indicators and applying the weight to the sum of differences. The weight is determined by determining a change over time of at least some of the historical physiological property indicators, applying a probability density function to the change to determine a probability density for the change occurring within a time period associated with the sensed physiological property indicators, and determining the weight based on the probability density. The forecast physiological property indicator includes a blood glucose value representing a forecast future blood glucose level, and the method further adjusts an insulin injection to correct blood glucose levels in response to the forecast future blood glucose levels.
Claims Coverage
The document includes four independent claims: clm-00001 (method), clm-00006 (system with forecaster device), clm-00009 (non-transitory computer readable medium), and clm-00010 (system using means-plus-function). Each independent claim covers the same core inventive workflow of patient-state classification combined with state-specific forecasting parameters derived from historical physiological property indicator sets, including selecting historical sets via differences and optional probability-density-based weighting, and producing a forecast that includes blood glucose and supports insulin injection adjustment.
Patient state classification with state-specific forecasting parameters
Applying at least one classification criterion to the contextual indicators to determine a patient state of a plurality of possible patient states, each patient state associated with a respective set of forecasting parameters, and applying the set of forecasting parameters associated with the determined patient state to sensed physiological property indicators to determine at least one forecast physiological property indicator representing a forecast physiological property of the patient at a future time.
Historical indicator-set selection using comparing and time-dependent function
Each set of forecasting parameters includes a plurality of sets of historical physiological property indicators, and applying the set comprises comparing the sensed patient physiological property indicators to each of the sets of historical physiological property indicators, selecting at least one of the sets based on said comparing, generating a time-dependent function representing the selected at least one set, and determining the at least one forecast physiological property indicator using the generated time-dependent function.
Probability-density-based weighting for change-over-time
Determining a sum of differences between the sensed patient physiological property indicators and respective ones of the historical physiological property indicators, determining a weight associated with the set of historical physiological property indicators and applying the weight to the determined sum of differences, wherein determining the weight comprises determining a change over time of at least some of the historical physiological property indicators, applying a probability density function to the change to determine a probability density for the change occurring within a time period associated with the sensed physiological property indicators, and determining the weight based on the determined probability density.
Forecast blood glucose value with insulin injection correction
Each of the at least one forecast physiological property indicators includes a blood glucose value representing a forecast future blood glucose level of the patient, and adjusting an insulin injection to correct blood glucose levels in response to the forecast future blood glucose levels indicated by the at least one forecast physiological property indicator.
Forecasting architecture using forecaster device and processing
A system comprising a forecaster device having at least one processor configured to receive sensed physiological property indicators and contextual indicators, apply at least one classification criterion to determine a patient state, and apply the set of forecasting parameters associated with the determined patient state to determine at least one forecast physiological property indicator, and a computer for managing insulin injection configured to receive the at least one forecast physiological property indicators and adjust an insulin injection in response to the forecast future blood glucose levels.
Non-transitory computer readable medium implementing forecasting and insulin adjustment
Codes which when executed cause at least one processor to receive signals representing sensed physiological property indicators, receive signals representing one or more contextual indicators, apply at least one classification criterion to determine a patient state associated with a respective set of forecasting parameters, apply the set to determine at least one forecast physiological property indicator at a future time, and cause a computer for managing insulin injection to adjust an insulin injection in response to forecast future blood glucose levels.
Across the independent claims, the inventive coverage centers on contextual-indicator classification into one of a plurality of patient states, applying state-associated forecasting parameters built from plurality sets of historical physiological property indicators, selecting historical sets using comparing based on sums of differences and probability-density-derived weighting from change-over-time, and generating a forecast that includes blood glucose and supports insulin injection adjustment to correct blood glucose levels in response to the forecast.
Stated Advantages
Facilitating physiological property forecasting for detecting disease complications.
Determining a forecast physiological property indicator representing a forecast physiological property of the patient at a future time based on patient state and historical physiological property indicators.
Correcting blood glucose levels by adjusting insulin injection in response to forecast future blood glucose levels.
Documented Applications
Detecting disease complications using physiological property forecasting based on sensed physiological property indicators and contextual indicators.
Forecasting blood glucose values for a patient and adjusting insulin injection to correct blood glucose levels in response to the forecast future blood glucose level.
Interested in licensing this patent?