updates to cou

exposure3_module/static/exposure_COU.html

@@ -72,16 +72,21 @@ literature. For polymer matrices that are not included in this list
 (“Other polymer”), CHRIS has two approaches. If the polymer glass
 transition temperature is not specified, an ultra-conservative diffusion
 coefficient that assumes the polymer has the properties of water is
-used. However, if the glass transition temperature is specified, a
+used. If the glass transition temperature is specified, a
 less-conservative upper bound diffusion coefficient is estimated using a
-machine learning method. Note that the worst-case diffusion coefficient
-is only defined over a molecular weight range of up to 1100 g/mol.
-Therefore, for substances with a molecular weight > 1100 g/mol, the
-value of the diffusion coefficient assuming a molecular weight of 1100
-g/mol is used as a conservative value. If the machine learning method is
-used to estimate the diffusion coefficient, a warning is issued if the
-polymer/solute combination falls outside the domain of the training
-data, indicating that the prediction may not be reliable.</p>
+machine learning method. For the listed polymeric systems and the
+ultra-conservative assumption, note that the worst-case diffusion
+coefficient is only defined over a molecular weight range of up to 1100
+g/mol. Therefore, for substances with a molecular weight &gt; 1100
+g/mol, the value of the diffusion coefficient assuming a molecular
+weight of 1100 g/mol is used as a conservative value. When the machine
+learning method is used, a check is conducted to verify that the
+polymer/solute combination falls inside the applicability domain of the
+model. If it falls outside that domain, CHRIS instead uses the
+ultra-conservative assumption. Additionally, note that polymers with
+high fractional free volume (FFV) should not be used with the machine
+learning method because diffusivity may be underestimated. Instead, the
+ultra-conservative assumption should be used.</p>
 <p>In the absence of adequate toxicological and exposure data for a
 chemical in a polymeric matrix, a toxicological risk assessment can be
 conducted for systemic biocompatibility endpoints by comparing the

exposure3_module/static/exposure_COU.md

@@ -29,6 +29,6 @@ CHRIS provides clinically relevant, yet still conservative, exposure dose estima
 1. The total amount of the chemical is present in dilute concentrations (<= 2 m/v %).
 1. Any particles/aggregates of the chemical present in the polymer are much smaller than the smallest component dimension (<= 50x).
 
-While these assumptions are typically valid for bulk additives and impurities in biostable polymers, users of CHRIS must confirm conformance to the underlying assumptions or provide supporting justification to ensure compliance for a given system.  Further, CHRIS only enables system specific exposure estimates for fifty-three (53) polymeric systems that are generally biostable (non-swelling and non-degrading).  To estimate chemical release based on the model, the diffusion coefficient of the chemical in the polymer matrix must be specified.  For the fifty-three (53) listed polymeric systems, a worst-case (upper bound) diffusion coefficient, as a function of molecular weight, has been established based on data from the literature.  For polymer matrices that are not included in this list ("Other polymer"), CHRIS has two approaches.  If the polymer glass transition temperature is not specified, an ultra-conservative diffusion coefficient that assumes the polymer has the properties of water is used.  However, if the glass transition temperature is specified, a less-conservative upper bound diffusion coefficient is estimated using a machine learning method.  Note that the worst-case diffusion coefficient is only defined over a molecular weight range of up to 1100 g/mol.  Therefore, for substances with a molecular weight > 1100 g/mol, the value of the diffusion coefficient assuming a molecular weight of 1100 g/mol is used as a conservative value.  If the machine learning method is used to estimate the diffusion coefficient, a warning is issued if the polymer/solute combination falls outside the domain of the training data, indicating that the prediction may not be reliable.
+While these assumptions are typically valid for bulk additives and impurities in biostable polymers, users of CHRIS must confirm conformance to the underlying assumptions or provide supporting justification to ensure compliance for a given system.  Further, CHRIS only enables system specific exposure estimates for fifty-three (53) polymeric systems that are generally biostable (non-swelling and non-degrading).  To estimate chemical release based on the model, the diffusion coefficient of the chemical in the polymer matrix must be specified.  For the fifty-three (53) listed polymeric systems, a worst-case (upper bound) diffusion coefficient, as a function of molecular weight, has been established based on data from the literature.  For polymer matrices that are not included in this list ("Other polymer"), CHRIS has two approaches.  If the polymer glass transition temperature is not specified, an ultra-conservative diffusion coefficient that assumes the polymer has the properties of water is used.  If the glass transition temperature is specified, a less-conservative upper bound diffusion coefficient is estimated using a machine learning method.  For the listed polymeric systems and the ultra-conservative assumption, note that the worst-case diffusion coefficient is only defined over a molecular weight range of up to 1100 g/mol.  Therefore, for substances with a molecular weight > 1100 g/mol, the value of the diffusion coefficient assuming a molecular weight of 1100 g/mol is used as a conservative value.  When the machine learning method is used, a check is conducted to verify that the polymer/solute combination falls inside the applicability domain of the model.  If it falls outside that domain, CHRIS instead uses the ultra-conservative assumption.  Additionally, note that polymers with high fractional free volume (FFV) should not be used with the machine learning method because diffusivity may be underestimated.  Instead, the ultra-conservative assumption should be used.
 
 In the absence of adequate toxicological and exposure data for a chemical in a polymeric matrix, a toxicological risk assessment can be conducted for systemic biocompatibility endpoints by comparing the exposure estimate to an appropriate threshold of toxicological concern (TTC).  This is the approach used by CHRIS in this module. The TTC values are based on systemic toxicity, thus CHRIS can address acute systemic toxicity, subacute/subchronic toxicity, genotoxicity, carcinogenicity, and reproductive and developmental toxicity.  It does not, however, address cytotoxicity, sensitization, irritation, hemocompatibility, material mediated pyrogenicity, or implantation. Therefore, an MOS >= 1 implies the chemical will not raise a safety concern with respect to only the systemic biocompatibility endpoints, provided the chemical is not within the cohort of concern, which is reflected in the output of CHRIS.