Adding sql queries that I need for the analysis

Scripts/Data_summaries.sql

@@ -0,0 +1,64 @@
+
+/* How many trials were included?
+ * How many trial were inspected?
+ * How many trials were reserved for downloaded?
+ * How many trials didn't get included for some technical reason?
+ * 
+********* Data from 2023-03-29 ***********
+Of Interest  1981
+Reserved    1709 #I believe this is lower than the downloaded number because I reserved them earlier
+Downloaded  1960
+Incomplete  3 #there were are few http 500 and 404 codes
+******************************************
+ * Note there were 21 missing trials of interest.
+ * */
+select status,count(distinct nct_id) from http.download_status ds 
+group by status;
+
+/* Get a list of trials
+ * -- There are currently 304 trials for which I was able to extract unique snapshots (2023-03-29)
+ * */
+select count(distinct nct_id) from history.trial_snapshots ts
+
+/* Get the number of listed conditions
+ * -- There are only 609 listed (MeSH classified) conditions from 284 trials
+ * I may need to expand how I address conditions
+ */
+select count(*)
+from ctgov.browse_conditions bc
+where
+mesh_type = 'mesh-list'
+and
+nct_id in (select distinct nct_id from history.trial_snapshots ts)
+;
+
+select count(distinct nct_id)
+from ctgov.browse_conditions bc
+where
+mesh_type = 'mesh-list'
+and
+nct_id in (select distinct nct_id from history.trial_snapshots ts)
+;
+
+/*
+ * If I were to expand that to non-coded conditions that would be
+ * 304 trials with 398 conditions
+ * */
+select count(distinct nct_id)
+from ctgov.conditions bc
+where
+nct_id in (select distinct nct_id from history.trial_snapshots ts)
+;
+select count(*) from ctgov.conditions c 
+where 
+nct_id in (select distinct nct_id from history.trial_snapshots ts)
+
+
+/* Get the number of matches from UMLS
+ * There are about 5,808 proposed matches.
+ * 
+ */
+select count(*) from "DiseaseBurden".trial_to_icd10 tti ;
+--1383 before run at 8pm 2023-03-29
+
+

Scripts/GroupingTrials.sql

@@ -0,0 +1,32 @@
+select * from "DiseaseBurden".icd10_to_cause itc ;
+select * from "DiseaseBurden".cause c ;
+select
+
+
+select c.id, count(distinct code)
+from "DiseaseBurden".cause c 
+    join "DiseaseBurden".icd10_to_cause itc 
+        on c.cause = itc.cause_text 
+group by c.id
+order by c.id
+;
+
+select count(distinct nct_id) from "DiseaseBurden".trial_to_icd10 tti 
+where tti.approved = 'accepted';
+
+select nct_id, "condition", ui
+from "DiseaseBurden".trial_to_icd10 tti 
+where tti.approved = 'accepted';
+
+
+select tti.nct_id, tti."condition",itc.cause_text 
+from "DiseaseBurden".trial_to_icd10 tti 
+    join "DiseaseBurden".icd10_to_cause itc
+        on replace(REPLACE(tti.ui,'-',''),'.','') = replace(REPLACE(itc.code ,'-',''),'.','')
+where 
+    tti.approved = 'accepted'
+    and 
+    itc.cause_text not in ('Non-communicable diseases','Neoplasms','Mental disorders','Other non-communicable diseases')
+group by tti.nct_id , tti."condition" , itc.cause_text 
+order by nct_id 
+;

scripts/Icd10ConditionsMatching/start.sh

@@ -0,0 +1 @@
+waitress-serve --port=5000 --call 'Icd10ConditionsMatching:create_app'