Method, system and computer readable medium for assessing actionable glycemic risk
A system, method and non-transient computer readable medium for assessing the opportunity to address either hyperglycemic or hypoglycemic risk (or both) in patients with diabetes based on historical continuous glucose monitoring (CGM) data.
1. A method for assessing actionable glycemic risk comprising:
obtaining an output continuous glucose monitoring (BG) signal at a time t;
determining a glycemic cost signal at the time t based on the output continuous glucose monitoring BGsignal with a glycemic cost function algorithm that attributes equal cost to each of two end points (BGtarget,lo, BGtarget,hi) of a desired blood glucose target range;
using the glycemic cost signal to quantify actionable risk and unaddressable risk by determining an instantaneous hypoglycemia risk at the time t and an instantaneous hyperglycemia risk at the time t based on the glycemic cost signal, wherein actionable risk includes actionable hypoglycemia risk and actionable hyperglycemia risk;
averaging each of the instantaneous hypoglycemia risk and the instantaneous hyperglycemia risk over a defined interval to determine a moving average hypoglycemia risk and a moving average hyperglycemia risk;
averaging each of the moving average hypoglycemia risk and the moving average hyperglycemia risk across multiple days to generate a hypoglycemia risk profile (lop(s)) and a hyperglycemia risk profile (hip(s)) at time s of the day;
administering insulin to a patient, via an insulin device, based on the generated hypoglycemia risk profile (lop(s)) and hyperglycemia risk profile (hip(s)).
2. The method of claim 1 , wherein the glycemic cost function algorithm is:
J
(
BG
)
=
c
[
ln
(
BG
BG
target
,
lo
·
BG
target
,
hi
)
]
2
,
where
c is a constant and BG is a blood glucose value.
3. The method of claim 1 , wherein the instantaneous hypoglycemia risk and the instantaneous hyperglycemia risk are determined by the algorithm:
rlo(BG)=J(BG), if BG<sqrt(BGtarget,lo*BGtarget,hi) 0, otherwise
rhi(BG)=J(BG), if BG>=sqrt(BGtarget,lo*BGtarget,hi) 0, otherwise, wherein
J(BG) is the glycemic cost function, rlo(BG) is the instantaneous hypoglycemia risk and rhi(BG) is the instantaneous hyperglycemia risk.
4. The method of claim 1 , comprising:
discounting the instantaneous hypoglycemia risk or the instantaneous hyperglycemia risk if the slope of the BG signal indicates that the BG signal is trending towards a midpoint of the blood glucose target range.
5. The method of claim 1 , comprising:
summing the hypoglycemia risk profile and the hyperglycemia risk profile to generate a total risk profile (trp).
6. The method of claim 5 , comprising:
using the hypoglycemia risk profile to generate an actionable hypoglycemia risk; and
using the hyperglycemia risk profile to generate an actionable hyperglycemia risk.
7. The method of claim 6 , wherein the actionable hypoglycemia risk is generated by the algorithm ARlow(s)=lop(s)*[lop(s)−hip(s)] + /trp(s), and the actionable hyperglycemia risk is generated by the algorithm ARhigh(s)=hip(s)*[hip(s)−lop(s)] + /trp(s).
8. The method of claim 7 , wherein the unaddressable risk is generated by the algorithm:
UR( s )=[trp( s )− AR low( s )− AR high( s )]/2.
9. The method of claim 1 , comprising:
interpolating the BG signal to fill gaps in the BG signal.
10. The method of claim 1 , comprising:
using the BG signal only for real-time determination of actionable risk and unaddressable risk.
11. The method of claim 1 , comprising:
using the BG signal for retrospective analysis of actionable risk and unaddressable risk.
12. A system for assessing actionable glycemic risk comprising:
a digital processor;
a continuous blood glucose monitoring (BG) sensor in communication with the digital processor, the continuous blood glucose monitoring (BG) sensor being configured to generate a blood glucose signal;
a risk determination module, configured to:
receive the glucose signal from the continuous glucose monitoring sensor, and generate an actionable hypoglycemia risk signal, an actionable hyperglycemia risk signal, and an unaddressable risk signal using a glycemic cost function algorithm that attributes equal cost to each of two end points (BGtarget,lo, BGtarget,hi) of a desired blood glucose target range,
determine an instantaneous hypoglycemia risk at the time t and an instantaneous hyperglycemia risk at the time t based on the risk signals,
average each of the instantaneous hypoglycemia risk and the instantaneous hyperglycemia risk over a defined interval to determine a moving average hypoglycemia risk and a moving average hyperglycemia risk;
average each of the moving average hypoglycemia risk and the moving average hyperglycemia risk across multiple days to generate a hypoglycemia risk profile (lop(s)) and a hyperglycemia risk profile (hip(s)) at time s of the day; and
an insulin device that administers insulin to a patient based on the generated hypoglycemia risk profile (lop(s)) and hyperglycemia risk profile (hip(s)).
13. The system of claim 12 , wherein the glycemic cost function algorithm is:
J
(
BG
)
=
c
[
ln
(
BG
BG
target
,
lo
·
BG
target
,
hi
)
]
2
,
where c is a constant and BG is a blood glucose value.
14. The system of claim 12 , wherein the risk determination module is configured to:
discount the instantaneous hypoglycemia risk or the instantaneous hyperglycemia risk when the slope of the BG signal indicates that the BG signal is trending towards a midpoint of the blood glucose target range.
15. The system of claim 12 , wherein the risk determination module is configured to:
sum the hypoglycemia risk profile and the hyperglycemia risk profile to generate a total risk profile (trp).
16. The system of claim 15 , wherein the risk determination module is configured to:
a. generate an actionable hypoglycemia risk signal based on the hypoglycemia risk profile by an algorithm ARlow(s)=lop(s)*[lop(s)−hip(s)] + /trp(s);
b. generate an actionable hyperglycemia risk signal based on the hyperglycemia risk profile by an algorithm ARhigh(s)=hip(s)*[hip(s)−lop(s)] + /trp(s); and
c. generate an unaddressable risk signal by an algorithm UR(s)=[trp(s)−ARlow(s)−ARhigh(s)]/2.
17. The system of claim 12 , comprising:
an insulin pump in communication with the digital processor and configured to dispense insulin in response to real-time analysis of at least one of the actionable hypoglycemia risk signal, the actionable hyperglycemia risk signal, and the unaddressable risk signal.