Author |
Message |
iyas (iyas)
Junior Member Username: iyas
Post Number: 6 Registered: 7-2007
| Posted on Wednesday, February 06, 2008 - 1:23 am: | |
thanks everybody for your kind help |
Cesar Guerrero (cesar)
Member Username: cesar
Post Number: 14 Registered: 3-2006
| Posted on Tuesday, February 05, 2008 - 10:16 am: | |
Hi, If you work with Bruker MPA, maybe you are able to work with the OPUS software. If yes, go to Evaluation, and then "calibration design". This will help you to make your mixtures avoiding correlations between drugs. Best regards, Cesar |
iyas (iyas)
New member Username: iyas
Post Number: 5 Registered: 7-2007
| Posted on Tuesday, February 05, 2008 - 9:55 am: | |
one of my colleauges has advised my be this please give me your opinion for this can you also give a professor contact who works with brucker my friend said : I can�t give you a solid answer as it would require much more time and information gathering, but a good estimate for a PLS model is that it doesn�t get anybetter over 60-80 samples total, so you can only play with the distribution of samples in this number, perhaps streaching it to 100 samples. Take 60 samples as a good starting point and prepare an experimental design in which you use the 20 samples of each of the 3 materials to spread of the variations systematically. Once done you have 60 samples in which everything varies. No take a PLS1 model and give the 20 samples of Sampletype A the arbitrary value 100 and the last 40 value of the two products not to beidentified the value 0, now run your PLS1 model with anticipated 6-7 factors (2+ for each product) and you should get a good prediction of the correct material of 80-120 and -10 - +10 for the bad once. For the other components swap places so Sample B becomes 100 and A+C is 0 |
iyas (iyas)
New member Username: iyas
Post Number: 4 Registered: 7-2007
| Posted on Tuesday, February 05, 2008 - 9:47 am: | |
one of my colleauges has advised my be this please give me your opinion for this can you also give aprofessor conatc hoiw works with brucker my friend said : I can�t give you a solid answer as it would require much more time and information gathering, but a good estimate for a PLS model is that it doesn�t get anybetter over 60-80 samples total, so you can only play with the distribution of samples in this number, perhaps streaching it to 100 samples. Take 60 samples as a good starting point and prepare an experimental design in which you use the 20 samples of each of the 3 materials to spread of the variations systematically. Once done you have 60 samples in which everything varies. No take a PLS1 model and give the 20 samples of Sampletype A the arbitrary value 100 and the last 40 value of the two products not to beidentified the value 0, now run your PLS1 model with anticipated 6-7 factors (2+ for each product) and you should get a good prediction of the correct material of 80-120 and -10 - +10 for the bad once. For the other components swap places so Sample B becomes 100 and A+C is 0. |
Ralf Marbach (ralf)
New member Username: ralf
Post Number: 3 Registered: 9-2007
| Posted on Tuesday, February 05, 2008 - 8:37 am: | |
BEWARE of "riding the cliff" --- given that statistical calibrations (PLS, PCR etc.) on tablet applications rarely achieve better than r=0.9, you will need >100 samples, better >200 to get precise PRESS values. Otherwise, you are just riding the cliff; see Journal of Biomedical Optics, Jan 2002, Volume 7, Issue 1, pp. 130-147. Ralf |
Bruce H. Campbell (campclan)
Moderator Username: campclan
Post Number: 110 Registered: 4-2001
| Posted on Tuesday, February 05, 2008 - 7:49 am: | |
Iyas, On an NIR basis I would recommend at least 40 calibration samples, each prepared in duplicate for a total of 80. The validation set size I recommend is at least 15, again prepared in duplicate for a total of 30. My reasoning for this number and the duplicates is you have at least four parameters, the three vitamins and the homogeniety problem. Further, scans should be made of the duplicates during one day if possible and then repeated at least a week later. You can then use the predictions (and possibly the calibrations done on each day's scans) in an ANOVA approach to examine the effect of homogeneity and time elapsed between making the various samples and the scanning of the same. Bruce |
iyas (iyas)
New member Username: iyas
Post Number: 3 Registered: 7-2007
| Posted on Tuesday, February 05, 2008 - 7:29 am: | |
i am doing the test on brucker device MPA and I also need to know how many samples i must do and the distribution of the tree components in the calibration and validation set and the problem that homogeneity of this powders take time by IPC to make an analysis to confirm the homogeneity of the powders to make the tabelting waiting for your answer iyas |
iyas (iyas)
New member Username: iyas
Post Number: 2 Registered: 7-2007
| Posted on Tuesday, February 05, 2008 - 7:24 am: | |
if you please I will ask you to help me if you could tell me how to do quantification test 1- measuring a pharmaceutical dosage form like B1, B6 , B12 vitamins granules so that to clarify the role of NIR in improving and saving time in the IPC (IN PROCESS CONTROL ) during mixing and homogeneity test , I like to know how many sample I must measure and the number of calibration and validation samples must I prepare for the dosage form of three active materials and for one material and i know your are very busy , but if you please it is too urgent to me I have searched all the web many studies made only one components if you have any contact of person who will help me and you can ask him to help me sorry again for my long question waiting for your answer iyas aldib p.s the formulation of this tablet granules the low dose is referred to the the B12 low dose in the B1 , B6 ,b12 vitamins preparations like tablets or ampoules like Composici�n: Cada ampolla de 3 ml contiene: Vitamina B1 (como clorhidrato de Tiamina) 100 mg; Vitamina B6 (como clorhidrato de Piridoxina) 100 mg; Vitamina B12 (Hidroxocobalamina) 10.000 mcg or Chaque comprim� enrob� de Vitamine Fort contient : Vitamine B1 100 mg. Vitamine B6 100 mg. Vitamine B12 1000 mcg. _____________________________________________________________________ |
David W. Hopkins (dhopkins)
Senior Member Username: dhopkins
Post Number: 132 Registered: 10-2002
| Posted on Monday, February 04, 2008 - 1:41 pm: | |
Hi Iyas, I don't think you will have any troubles distinguishing these 3 vitamins, the problems will be at what levels do they need to be measured? Also, what excipient(s) are present in the mix, and at what concentrations? I would recommend that you start with feasibility studies just preparing the samples with carefully prepared concentrations gravimetically. Then you can determine whether you can actually measure the vitamins at the target levels. Then you could do a designed experiment to make calibrations for the mixture of the 3 vitamins. I agree with Deepak that I think 2nd Derivatives may be useful with MSC (after 2Der). However, I think that just MSC on the original spectra may be easier to interpret, and possibly as accurate, so I would recommend trying both procedures. Please let us know what the target concentrations of actives and excipients are. My expectation is that the B12 may be at very low levels, and you will just have to see whether you can detect it at that level. But you will need to investigate the level of the others too. Best regards, Dave |
Deepak Sharma (deepak)
New member Username: deepak
Post Number: 3 Registered: 11-2007
| Posted on Monday, February 04, 2008 - 12:37 pm: | |
Exactly, i don't know , what type of instrument and software you are using for this activity. If i am correct, you have to prepare a range of synthetic samples(depending on your release limits), having different concentration of these three active drugs vitamins b1-b6 b12. Get the actual concentration of vitamin b1-b6-b12,by analysing samples on primary method.Plot a linear calibration curve between referee method and nir method, and get the quantitative results by NIR. I will prefer 2nd Derivative, but it all depends on your raw data. Hope this will serve a little bit for you. regards |
iyas (iyas)
New member Username: iyas
Post Number: 1 Registered: 7-2007
| Posted on Monday, February 04, 2008 - 11:30 am: | |
if you please i need help in making quantative assy for a tablet conatining of 3 active drugs like vitamins like b1, b6 , b12 how to do teh method and the calibration and the validation set , must i use the snv and the firt and second deivative should i do the powder and the tablet sepratley and how musch of smples and batchs must ido wiatingf royour help all te aricles in the internet is delaing with two componenets iam pharacmeutical analysis msc student kind regrads and waiting fo your help and answeers iyas |