turn on suggestions

Auto-suggest helps you quickly narrow down your search results by suggesting possible matches as you type.

Showing results for

- JMP User Community
- :
- Discussions
- :
- tablet dissolution profile

Topic Options

- Subscribe to RSS Feed
- Mark Topic as New
- Mark Topic as Read
- Float this Topic for Current User
- Bookmark
- Subscribe
- Printer Friendly Page

Highlighted

- Mark as New
- Bookmark
- Subscribe
- Subscribe to RSS Feed
- Get Direct Link
- Email to a Friend
- Report Inappropriate Content

A week ago
(181 views)

Hi,

I have a set of tablets in which one component determines how slowly the active is released (measured at 1, 2, 3, 4, 5 and 6hours).

If I have the results for tablets containing 10, 15, 20 and 25% of this component how can I create a model that will tell me how much of that component is needed to obtain a target profile (target release values at target timepoints)?

Thanks

1 ACCEPTED SOLUTION

Accepted Solutions

- Mark as New
- Bookmark
- Subscribe
- Subscribe to RSS Feed
- Get Direct Link
- Email to a Friend
- Report Inappropriate Content

A week ago
(215 views)

Solution

Also, you might try saving the model and creating a new response to optimize. A good choice would be the root mean square error (RMSE). A column formula could compare the predicted response with the target response using something like this

```
Sqrt(
Sum(
(P(t1,X) - T(t1))^2 +
(P(t2,X) - T(t2))^2 +
(P(t3,X) - T(t3))^2 +
(P(t4,X) - T(t4))^2 +
(P(t5,X) - T(t5))^2
)
)
```

where the P(t,X) is the fitted model of the predicted response at time t and excipient level X. T(t) is the target level at time t. You could use the Graph > Profiler with this column to optimize X (minimize response with desirability function).

Learn it once, use it forever!

7 REPLIES

- Mark as New
- Bookmark
- Subscribe
- Subscribe to RSS Feed
- Get Direct Link
- Email to a Friend
- Report Inappropriate Content

A week ago
(176 views)

One JMP platform you may want to take a look at is the Fit Curve platform for the type of problem you are trying to address. Here is the link to the JMP online documentation for the Fit Curve platform.

https://www.jmp.com/support/help/13-2/Fit_Curve.shtml

- Mark as New
- Bookmark
- Subscribe
- Subscribe to RSS Feed
- Get Direct Link
- Email to a Friend
- Report Inappropriate Content

A week ago
(166 views)

- Mark as New
- Bookmark
- Subscribe
- Subscribe to RSS Feed
- Get Direct Link
- Email to a Friend
- Report Inappropriate Content

A week ago
(153 views)

- Mark as New
- Bookmark
- Subscribe
- Subscribe to RSS Feed
- Get Direct Link
- Email to a Friend
- Report Inappropriate Content

A week ago
(148 views)

You might also try using Analyze > Reliability > Degradation. Use the excipient level as X. You can try transforms on Y and X with a linear path or use a nonlinear path.

Learn it once, use it forever!

- Mark as New
- Bookmark
- Subscribe
- Subscribe to RSS Feed
- Get Direct Link
- Email to a Friend
- Report Inappropriate Content

A week ago
(132 views)

I probably do not fully understand how the degradation model works but...

When I have a target profile it means that I have target values for the various timepoints (e.g. timepoint 100: 10%, 200: 15%, 400: 45%, etc.) so not just one point. And I would need a model that would tell me what concnetration of X would give me a closest match.

My best shot so far was to create a single factor DOE where I set the timepoints as the responses and simply overwrote the parameters in the provided table with the parameters and results of my experiments and used the profilers.

But there must be a better way to do that, I guess.

- Mark as New
- Bookmark
- Subscribe
- Subscribe to RSS Feed
- Get Direct Link
- Email to a Friend
- Report Inappropriate Content

A week ago
(119 views)

Please don't be put off by the terminology. The 'degradation path' in the Degradation platform is just a model of the change in the response (dissolution) over time under the influence of a factor (excipient level). You can then profile the response under different conditions to find the target profile. You can use inverse prediction to find the condition that achieves a desired response level, too.

Learn it once, use it forever!

- Mark as New
- Bookmark
- Subscribe
- Subscribe to RSS Feed
- Get Direct Link
- Email to a Friend
- Report Inappropriate Content

A week ago
(216 views)

Also, you might try saving the model and creating a new response to optimize. A good choice would be the root mean square error (RMSE). A column formula could compare the predicted response with the target response using something like this

```
Sqrt(
Sum(
(P(t1,X) - T(t1))^2 +
(P(t2,X) - T(t2))^2 +
(P(t3,X) - T(t3))^2 +
(P(t4,X) - T(t4))^2 +
(P(t5,X) - T(t5))^2
)
)
```

where the P(t,X) is the fitted model of the predicted response at time t and excipient level X. T(t) is the target level at time t. You could use the Graph > Profiler with this column to optimize X (minimize response with desirability function).

Learn it once, use it forever!