|
Post by dreamboatcruise on Jun 29, 2016 11:09:55 GMT -5
I'm sure you are correct that this wouldn't be able to predict metabolic behavior of particular patients. Do you know that it uses a "standardized meal"? Doesn't seem like it would be too difficult to add in some basic model of digestion/absorption to broadly capture much of the variability across the population (perhaps missing some statistical outliers) and then allow various simulated meals (specified by size and nutritional content or by glycemic index) to be fed to the various simulated patients. I went back and read the paper this time rather than just skim it! The algorithm used seems fairly robust, it's derived from the AP algorithms and FDA approved for modelling (but not for human use). The PK curve is taken from the NCT01544881 (MKC-TI-177) trial back in 2013 which set the PK curve for the Afrezza prescribing note. They then took individual variances within the trial data and used that to create a much large data set (the wonders of MATLAB). This gave them a 100 virtual humans based of the characteristics of the original 12 real people. They feed the 100 virtual humans an ISO standard 50g meal and that's where the results came from. As you correctly said this bypasses all the nasty real world issues. It is based on the clamp data and not real food. This is why you would not be wise to use it medically. It fails the pizza test! Wish I had more time to dig into this. I've previously worked in the field of computer modeling and simulation... though not physiological. To me this is certainly a very interesting and potentially hugely beneficial technology. I wasn't even aware that the FDA had in place a process of approving simulation models. I'm very curious as to what criteria they use to approve and exactly what it is they are approving it to be used for. Are they moving towards eventually accepting these simulations at least as augmentation in decisions about drug approval and/or labeling? I would assume that there is some overlap in this modeling with the artificial pancreas effort, and according to recent reports Minimed has an artificial pancreas device likely to be approved by next year. Exciting, fascinating stuff in the world of medicine.
|
|
|
Post by agedhippie on Jun 30, 2016 7:44:28 GMT -5
I went back and read the paper this time rather than just skim it! The algorithm used seems fairly robust, it's derived from the AP algorithms and FDA approved for modelling (but not for human use). The PK curve is taken from the NCT01544881 (MKC-TI-177) trial back in 2013 which set the PK curve for the Afrezza prescribing note. They then took individual variances within the trial data and used that to create a much large data set (the wonders of MATLAB). This gave them a 100 virtual humans based of the characteristics of the original 12 real people. They feed the 100 virtual humans an ISO standard 50g meal and that's where the results came from. As you correctly said this bypasses all the nasty real world issues. It is based on the clamp data and not real food. This is why you would not be wise to use it medically. It fails the pizza test! Wish I had more time to dig into this. I've previously worked in the field of computer modeling and simulation... though not physiological. To me this is certainly a very interesting and potentially hugely beneficial technology. I wasn't even aware that the FDA had in place a process of approving simulation models. I'm very curious as to what criteria they use to approve and exactly what it is they are approving it to be used for. Are they moving towards eventually accepting these simulations at least as augmentation in decisions about drug approval and/or labeling? I would assume that there is some overlap in this modeling with the artificial pancreas effort, and according to recent reports Minimed has an artificial pancreas device likely to be approved by next year. Exciting, fascinating stuff in the world of medicine. I must admit until I read that paper I wasn't aware of FDA approved models either. I have never seen a drug or device approval purely based on a model (or even partially that I was aware of) so I suspect that they allow the data as background. In this case I could see them allowing the model data as justification as to why you could dose later than normal. Without the justification they would likely not allow late dosing based on safety grounds since it goes against current evidence based practice. I would love to look at the model they are using. Physiological models are difficult to pull off because the level of complexity is almost infinite - how finely do you want to model all the factors and interactions. Since this is the AP test model it must be fairly complex and I am curious to see what level they go to.
|
|
|
Post by peppy on Jun 30, 2016 8:16:24 GMT -5
Wish I had more time to dig into this. I've previously worked in the field of computer modeling and simulation... though not physiological. To me this is certainly a very interesting and potentially hugely beneficial technology. I wasn't even aware that the FDA had in place a process of approving simulation models. I'm very curious as to what criteria they use to approve and exactly what it is they are approving it to be used for. Are they moving towards eventually accepting these simulations at least as augmentation in decisions about drug approval and/or labeling? I would assume that there is some overlap in this modeling with the artificial pancreas effort, and according to recent reports Minimed has an artificial pancreas device likely to be approved by next year. Exciting, fascinating stuff in the world of medicine. I must admit until I read that paper I wasn't aware of FDA approved models either. I have never seen a drug or device approval purely based on a model (or even partially that I was aware of) so I suspect that they allow the data as background. In this case I could see them allowing the model data as justification as to why you could dose later than normal. Without the justification they would likely not allow late dosing based on safety grounds since it goes against current evidence based practice.
I would love to look at the model they are using. Physiological models are difficult to pull off because the level of complexity is almost infinite - how finely do you want to model all the factors and interactions. Since this is the AP test model it must be fairly complex and I am curious to see what level they go to. The type one diabetic I knew, gave herself subq insulin after meals. In the elevator. 80's, 90's, 2000 Looking back and wondering why, Safety in that the food had been ingested? More risk in taking the insulin and then not getting the food? Just reporting the subq insulin time table she used.
The least risk would seem to be, take the insulin while eating.
|
|
|
Post by agedhippie on Jun 30, 2016 12:51:25 GMT -5
I must admit until I read that paper I wasn't aware of FDA approved models either. I have never seen a drug or device approval purely based on a model (or even partially that I was aware of) so I suspect that they allow the data as background. In this case I could see them allowing the model data as justification as to why you could dose later than normal. Without the justification they would likely not allow late dosing based on safety grounds since it goes against current evidence based practice.
I would love to look at the model they are using. Physiological models are difficult to pull off because the level of complexity is almost infinite - how finely do you want to model all the factors and interactions. Since this is the AP test model it must be fairly complex and I am curious to see what level they go to. The type one diabetic I knew, gave herself subq insulin after meals. In the elevator. 80's, 90's, 2000 Looking back and wondering why, Safety in that the food had been ingested? More risk in taking the insulin and then not getting the food? Just reporting the subq insulin time table she used.
The least risk would seem to be, take the insulin while eating.
I do that as well and get slapped by other diabetics for doing it. The correct approach is to bolus for the minimum you expect to eat beforehand and then clean up the extra afterwards. Sadly I am idle so I tend to skip the first bolus and move straight to the clean up which gives worse results. For the most part I bolus properly at home and badly when I am out.
|
|
|
Post by dreamboatcruise on Jun 30, 2016 15:20:23 GMT -5
The type one diabetic I knew, gave herself subq insulin after meals. In the elevator. 80's, 90's, 2000 Looking back and wondering why, Safety in that the food had been ingested? More risk in taking the insulin and then not getting the food? Just reporting the subq insulin time table she used.
The least risk would seem to be, take the insulin while eating.
I do that as well and get slapped by other diabetics for doing it. The correct approach is to bolus for the minimum you expect to eat beforehand and then clean up the extra afterwards. Sadly I am idle so I tend to skip the first bolus and move straight to the clean up which gives worse results. For the most part I bolus properly at home and badly when I am out. Seems like you need a faster acting insulin. I've heard rumors there is one on the market
|
|
Deleted
Deleted Member
Posts: 0
|
Post by Deleted on Aug 10, 2016 19:19:19 GMT -5
|
|
|
Post by mannmade on Aug 10, 2016 19:28:24 GMT -5
Sanofi. At least as of a few months ago. I spoke with him at their HQ by phone.
|
|
Deleted
Deleted Member
Posts: 0
|
Post by Deleted on Aug 10, 2016 19:40:15 GMT -5
Sanofi. At least as of a few months ago. I spoke with him at their HQ by phone. So is Mannkind and Sanofi still has something going on?
|
|
|
Post by mannmade on Aug 10, 2016 19:54:46 GMT -5
No I was trying to help a friend get samples during transition and was referred to Sanofi by mnkd.
|
|
Deleted
Deleted Member
Posts: 0
|
Post by Deleted on Aug 10, 2016 20:34:01 GMT -5
No I was trying to help a friend get samples during transition and was referred to Sanofi by mnkd. Did they just transfer you to him automatically? Which HQ? When did he start working for Sanofi?
|
|
|
Post by mannmade on Aug 10, 2016 23:23:41 GMT -5
Sanofi HQ I asked for him with a reference from a friend who is a doctor and knows him.
|
|