Tests to perform on JMP for statistical analysis of my research data

camillebreu

Hello everyone! I need some advice.

For my master's thesis at the University Hospital of Antwerp, I am working on an experiment that measures how much and what plasticizers (= toxic plastic components) are released from plastic medical devices during medical use. Specifically, we measured the release of plasticizers from an infusion circuit during infusion of nutritional fluid at different time points.

 

In brief:

1. We infuse a nutritive fluid through a medical infusion circuit, capture it back at the end of the circuit at different time points (before infusion, after 12, 24 and 48 hours). We then measure the concentration of different plasticizers in the collected nutritional fluids. In this way, the leaching over time can be examined.
2. We performed this experiment on 3 different nutritive fluids, to see if there is a significant difference in the release of plasticizers between these fluids.
   1. For nutritive fluid 1, the experiment is performed in triplicate, for the others only once.

 

My database looks as follows:

 

Experiment	Nutritive fluid	Time points of collection (hours)	Concentration plasticizer 1 (micrograms)	Concentration plasticizer 2 (micrograms)	Concentration plasticizer 3 (micrograms)
1	fluid 1	0	x	x	X
1	fluid 1	12	x	x	x
1	fluid 1	24	x	x	x
1	fluid 1	48	x	x	x
2	fluid 2	0	x	x	X
2	fluid 2	12	x	x	x
2	fluid 2	24	x	x	x
2	fluid 2	48	x	x	x
3	fluid 3	0	x	x	X
3	fluid 3	12	x	x	x
3	fluid 3	24	x	x	x
3	fluid 3	48	x	x	x

 

(We are still waiting on the results of the concentration of the different plasticizers)

 

My question: which test(s) would be best to run on JMP for my statistical analysis? Or to show (visually) the significance/results of my data? In order to

1. Show the concentrations of plasticizer release. 
2. Track the plasticizer release over time.
3. To check the difference in plasticizer release between the 3 different nutritive fluids. 

 

Many thanks in advance! 

 

Kind regards,

Camille

Georg

There is no easy and single answer to this, especially w/o knowing the process, noise, physics and data in advance.

Enclosed you find a script that generates a table with two scripts and some modeled data.

A first approach for me would always be a visual judgement (by Graph Builder etc.).

A second approach could be the fit model platform, in this case least squares, see second script. The p-value may show you about significance of the parameter, fluid 3 is different for plasticizer 3, but it is difficult to find. With more data it would be easier to find. By pressing "?" and clicking the mouse in the report, you'll find context help to understand the details.

Good Luck!

 

 

 

Names Default To Here( 1 );

Random Reset( 42 );

db = New Table( "Plasticizers",
	add rows( 12 ),
	New Column( "Experiment", "Numeric", "Ordinal", set values( [1, 1, 1, 1, 2, 2, 2, 2, 3, 3, 3, 3] ) ),
	New Column( "Nutritive fluid",
		"Character",
		"Nominal",
		set values(
			{"fluid 1", "fluid 1", "fluid 1", "fluid 1", "fluid 2", "fluid 2", "fluid 2", "fluid 2", "fluid 3", "fluid 3", "fluid 3", "fluid 3"}
		)
	),
	New Column( "Time points of collection (hours)",
		Numeric,
		Continuous,
		set format( "Best", 12 ),
		set values( [0, 12, 24, 48, 0, 12, 24, 48, 0, 12, 24, 48] )
	),
	New Column( "Concentration plasticizer 1 (micrograms)",
		"Numeric",
		Continuous,
		set formula( Random Normal( 1, 1 ) + 0.1 * Name( "Time points of collection (hours)" ) )
	),
	New Column( "Concentration plasticizer 2 (micrograms)",
		"Numeric",
		Continuous,
		set formula( Random Normal( 3, 1 ) + 0.2 * Name( "Time points of collection (hours)" ) )
	),
	New Column( "Concentration plasticizer 3 (micrograms)",
		"Numeric",
		Continuous,
		set formula( Random Normal( 10, 1 ) + If( Name( "Nutritive fluid" ) == "fluid 3", 0.5, 0.1 ) * Name( "Time points of collection (hours)" ) )
	), 

);

db << Add Properties to Table(
	{New Script(
		"Concentration plasticizer 1 (micrograms) & 2 more vs. Time points of collection (hours)",
		Graph Builder(
			Size( 855, 739 ),
			Variables(
				X( :Name( "Time points of collection (hours)" ) ),
				Y( :Name( "Concentration plasticizer 1 (micrograms)" ) ),
				Y( :Name( "Concentration plasticizer 2 (micrograms)" ), Position( 1 ) ),
				Y( :Name( "Concentration plasticizer 3 (micrograms)" ), Position( 1 ) ),
				Group X( :Nutritive fluid )
			),
			Elements( Points( X, Y( 1 ), Y( 2 ), Y( 3 ), Legend( 14 ) ), Smoother( X, Y( 1 ), Y( 2 ), Y( 3 ), Legend( 15 ) ) )
		)
	), New Script(
		"Fit Least Squares",
		Fit Model(
			Y(
				:Name( "Concentration plasticizer 1 (micrograms)" ),
				:Name( "Concentration plasticizer 2 (micrograms)" ),
				:Name( "Concentration plasticizer 3 (micrograms)" )
			),
			Effects( :Nutritive fluid, :Name( "Time points of collection (hours)" ) ),
			Personality( "Standard Least Squares" ),
			Emphasis( "Effect Leverage" ),
			Run(
				:Name( "Concentration plasticizer 1 (micrograms)" ) << {Summary of Fit( 1 ), Analysis of Variance( 1 ), Parameter Estimates( 1 ),
				Lack of Fit( 0 ), Scaled Estimates( 0 ), Plot Actual by Predicted( 1 ), Plot Residual by Predicted( 1 ),
				Plot Studentized Residuals( 0 ), Plot Effect Leverage( 1 ), Plot Residual by Normal Quantiles( 0 ), Box Cox Y Transformation( 0 )},
				:Name( "Concentration plasticizer 2 (micrograms)" ) << {Summary of Fit( 1 ), Analysis of Variance( 1 ), Parameter Estimates( 1 ),
				Lack of Fit( 0 ), Scaled Estimates( 0 ), Plot Actual by Predicted( 1 ), Plot Residual by Predicted( 1 ),
				Plot Studentized Residuals( 0 ), Plot Effect Leverage( 1 ), Plot Residual by Normal Quantiles( 0 ), Box Cox Y Transformation( 0 )},
				:Name( "Concentration plasticizer 3 (micrograms)" ) << {Summary of Fit( 1 ), Analysis of Variance( 1 ), Parameter Estimates( 1 ),
				Lack of Fit( 0 ), Scaled Estimates( 0 ), Plot Actual by Predicted( 1 ), Plot Residual by Predicted( 1 ),
				Plot Studentized Residuals( 0 ), Plot Effect Leverage( 1 ), Plot Residual by Normal Quantiles( 0 ), Box Cox Y Transformation( 0 )}
			)
		)
	
	)}
);