You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
Copy file name to clipboardExpand all lines: articles/azure-health-insights/radiology-insights/transparency-note.md
+10-10Lines changed: 10 additions & 10 deletions
Display the source diff
Display the rich diff
Original file line number
Diff line number
Diff line change
@@ -79,7 +79,7 @@ The Radiology Insight adds several types of inferences/annotations to the origin
79
79
80
80
Example of a Clinical Radiology document with inferences:
81
81
82
-
](../media/radiology-insights/radiology-with-inferences.png#lightbox)
82
+
[](../media/radiology-insights/radiology-with-inferences.png#lightbox)
83
83
84
84
### Functional description of the inferences in scope and examples
85
85
@@ -95,7 +95,7 @@ Age mismatches are identified based on comparison of available Patient age infor
95
95
Sex mismatches identification is based on a comparison of the patient sex information within patient’s demographic meta-data on the one hand and on the other hand patient references, sex specific findings, and sex specific body parts in the text.
96
96
Conflicting sex information are tagged in the text.
97
97
98
-
](../media/radiology-insights/radiology-sex-mismatch.png#lightbox)
98
+
[](../media/radiology-insights/radiology-sex-mismatch.png#lightbox)
99
99
100
100
101
101
**Laterality Discrepancy**
@@ -104,23 +104,23 @@ A laterality, defined as "Left" (Lt, lft) and "Right" (rt, rght), along with an
104
104
No Mismatches on past content.
105
105
If only Laterality and no Anatomy is available in the Procedure Description, all opposite laterality in the text is tagged. For example: "left views" in Procedure Description will list all "right" words in the report text.
106
106
107
-
](../media/radiology-insights/radiology-laterality-mismatch.png#lightbox)
107
+
[](../media/radiology-insights/radiology-laterality-mismatch.png#lightbox)
108
108
109
109
110
110
**CompleteOrder Discrepancy**
111
111
112
112
Completeness mismatches can be made if the ordered procedure is an ultrasound for the ABDOMEN, RETROPERITONEAL, PELVIS, or US BREAST.
113
113
A completeness mismatch is made if either the order is complete and the text isn't, or vice versa.
114
114
115
-
](../media/radiology-insights/radiology-completeness-mismatch.png#lightbox)
115
+
[](../media/radiology-insights/radiology-completeness-mismatch.png#lightbox)
116
116
117
117
118
118
**LimitedOrder Discrepancy**
119
119
120
120
Completeness mismatches can be made if the ordered procedure is an ultrasound for the ABDOMEN, RETROPERITONEAL, PELVIS, or US BREAST.
121
121
A completeness mismatch is made if either the order is complete and the text isn't, or vice versa.
122
122
123
-
](../media/radiology-insights/radiology-completeness-limited-mismatch.png#lightbox)
123
+
[](../media/radiology-insights/radiology-completeness-limited-mismatch.png#lightbox)
124
124
125
125
126
126
**Finding**
@@ -139,7 +139,7 @@ Only Identifies Critical Result in the report text (not in meta-data)
139
139
The terms are based on Mass Coalition for the Prevention of Medical Errors:
](../media/radiology-insights/radiology-critical-result.png#lightbox)
142
+
[](../media/radiology-insights/radiology-critical-result.png#lightbox)
143
143
144
144
145
145
**FollowupRecommendation**
@@ -154,7 +154,7 @@ Cfr Sex Mismatch example
154
154
Radiology Insights extracts information such as modality, body part, laterality, view, and contrast from the procedure order. Ordered procedures are normalized using the Loinc codes using the LOINC/RSNA Radiology Playbook that is developed and maintained by the LOINC/RadLex Committee:
](../media/radiology-insights/radiology-procedure.png#lightbox)
157
+
[](../media/radiology-insights/radiology-procedure.png#lightbox)
158
158
159
159
160
160
**Communication**
@@ -189,7 +189,7 @@ Additionally, the sample documents in the impression section:
189
189
The observation in the right upper lobe will also be surfaced by the model as present guidance values. However, the model will also surface guidance values that are missed and not documented in the report. If these values are documented, the model could propose other candidate recommendations according to the clinical guideline for Pulmonary Nodule. This allows the radiologist to propose a candidate recommendation and/or enhance the documentation to include additional recommendations as part of the options.
190
190
191
191
192
-
](../media/radiology-insights/radiology-clinical-guidance.png#lightbox)
192
+
[](../media/radiology-insights/radiology-clinical-guidance.png#lightbox)
193
193
194
194
195
195
From the finding in the findings section (blue), the model proposes two recommendation candidates: Positron Emission Tomography and Imaging Guided Biopsy, both of the Lung. The guidance values surfaced for this finding are for the Lobe, Right Upper Lobe and for the Size, 12 mm. From the finding in the Impression section, the guidance value for the Lobe is returned, but the Size is missing and hence returned as a missing guidance value. No recommendations candidates are proposed.
@@ -238,7 +238,7 @@ In the sample below, a CVC or Central Venous Catheter needs to be inserted into
238
238
In the case of missing criteria, the documentation could be updated to also cover for the missing documented but executed criteria or a retrospect could be organized to understand why these important criteria weren't included in the procedure executed.
239
239
240
240
241
-
](../media/radiology-insights/radiology-quality-measure.png#lightbox)
241
+
[](../media/radiology-insights/radiology-quality-measure.png#lightbox)
242
242
243
243
244
244
Figure Quality Measures: The model reports that for the Quality Measure CVC INSERTION, the documentation doesn't comply with its quality criteria and classifies the type of compliance as "performance not met" in the model response. There's missing information regarding the use of a sterile probe cover.
@@ -287,7 +287,7 @@ The Model surfaces and highlights Risks, Scorings and Assessments with classific
287
287
In the sample below the model identifies two inferences: the ASCVD (Atherosclerotic Cardiovascular Disease) risk with value 17.6% and the Agatston Score (measure used to quantify the amount of calcium in the coronary arteries) with value zero. The 10-year ASCVD (Atherosclerotic Cardiovascular Disease) risk score is an estimate of the likelihood that a person will experience a cardiovascular event (like a heart attack or stroke) within the next 10 years. This score is calculated (not by the model) based on various risk factors, including age, cholesterol levels, blood pressure, smoking status, diabetes, and others. In this case, a risk score of 17.6% means that, based on the person's risk factors, there's a 17.6% chance of experiencing a cardiovascular event in the next 10 years.
288
288
289
289
290
-
](../media/radiology-insights/radiology-scoring-assessment.png#lightbox)
290
+
[](../media/radiology-insights/radiology-scoring-assessment.png#lightbox)
291
291
292
292
Figure Risk Scoring and Assessment: The model identifies two Scoring and Assessment instances (blue), one of category ASCVD Risk with value 17.6% and one of category Calcium Score with value 0.
0 commit comments