def build_dict():
d = {}
for effect in BACK_CALCULATABLE_DIAGNOSTIC_EFFECTS:
est,lower,upper = self.ma_unit.get_effect_and_ci(effect,
self.group_str,
self.mult)
conv_to_disp_scale = lambda x: meta_py_r.diagnostic_convert_scale(x, effect, convert_to="display.scale")
d_est,d_lower,d_upper = [conv_to_disp_scale(x) for x in [est,lower,upper]]
for i,Rsubkey in enumerate(["",".lb",".ub"]):
try:
d["%s%s" % (effect.lower(), Rsubkey)] = float([d_est,d_lower,d_upper][i])
except:
pass
x = self.getTotalSubjects()
d["total"] = float(x) if is_a_float(x) else None
x = self.prevalence_txt_box.text()
d["prev"] = float(x) if is_a_float(x) else None
d["conf.level"] = self.global_conf_level
# now grab the raw data, if available
d.update(self.get_raw_diag_data())
return d
def val_changed(self, val_str):
# Backup form state
old_ma_unit, old_table = self._save_ma_unit_and_table_state(
table = self.two_by_two_table,
ma_unit = self.ma_unit,
use_old_value=False)
old_prevalence = self.current_prevalence
new_text = self._get_txt_from_val_str(val_str)
no_errors, display_scale_val = self._text_box_value_is_between_bounds(val_str, new_text)
if no_errors is False: # There are errors
self.restore_ma_unit_and_table(old_ma_unit,old_table, old_prevalence)
calc_fncs.block_signals(self.entry_widgets, True)
if val_str == "est":
self.effect_txt_box.setFocus()
elif val_str == "lower":
self.low_txt_box.setFocus()
elif val_str == "upper":
self.high_txt_box.setFocus()
elif val_str == "prevalence":
self.prevalence_txt_box.setFocus()
calc_fncs.block_signals(self.entry_widgets, False)
return
# If we got to this point it means everything is ok so far
try:
if display_scale_val not in EMPTY_VALS:
display_scale_val = float(display_scale_val)
else:
display_scale_val = None
except ValueError:
# a number wasn't entered; ignore
# should probably clear out the box here, too.
print "fail."
return None
calc_scale_val = meta_py_r.diagnostic_convert_scale(display_scale_val,
self.cur_effect, convert_to="calc.scale")
if val_str == "est":
self.ma_unit.set_effect(self.cur_effect, self.group_str, calc_scale_val)
elif val_str == "lower":
self.ma_unit.set_lower(self.cur_effect, self.group_str, calc_scale_val)
elif val_str == "upper":
self.ma_unit.set_upper(self.cur_effect, self.group_str, calc_scale_val)
elif val_str == "prevalence":
pass
new_ma_unit, new_table = self._save_ma_unit_and_table_state(
table = self.two_by_two_table, ma_unit = self.ma_unit,
use_old_value=False)
new_prevalence = self._get_prevalence_str()
restore_old_f = lambda: self.restore_ma_unit_and_table(old_ma_unit, old_table, old_prevalence)
restore_new_f = lambda: self.restore_ma_unit_and_table(new_ma_unit, new_table, new_prevalence)
command = calc_fncs.CommandFieldChanged(restore_new_f=restore_new_f, restore_old_f=restore_old_f, parent=self)
self.undoStack.push(command)
self.current_prevalence = new_prevalence
def f(study_a, study_b):
ma_unit_A = study_a.get_ma_unit(outcome_name,follow_up)
ma_unit_B = study_b.get_ma_unit(outcome_name,follow_up)
if outcome_type is BINARY:
to_display_scale = lambda x: meta_py_r.binary_convert_scale(x, current_effect, convert_to="display.scale")
elif outcome_type is CONTINUOUS:
to_display_scale = lambda x: meta_py_r.continuous_convert_scale(x, current_effect, convert_to="display.scale")
elif outcome_type is DIAGNOSTIC:
to_display_scale = lambda x: meta_py_r.diagnostic_convert_scale(x, current_effect, convert_to="display.scale")
if outcome_type in (BINARY,CONTINUOUS):
outcome_data_A = ma_unit_A.get_effect_and_ci(current_effect, group_str, mult)
outcome_data_B = ma_unit_B.get_effect_and_ci(current_effect, group_str, mult)
outcome_data_A = [to_display_scale(c_val) for c_val in outcome_data_A]
outcome_data_B = [to_display_scale(c_val) for c_val in outcome_data_B]
elif outcome_type == DIAGNOSTIC:
# /\/\/\/\
(outcome_data_A, outcome_data_B) = ([],[])
# |
# \_____/
for diag_metric in ["Sens","Spec"]: # this order corresponds to the order displayed on the spreadsheet
est_and_ci_A = ma_unit_A.get_effect_and_ci(diag_metric, group_str, mult)
est_and_ci_B = ma_unit_B.get_effect_and_ci(diag_metric, group_str, mult)
est_and_ci_A = [to_display_scale(c_val) for c_val in est_and_ci_A]
est_and_ci_B = [to_display_scale(c_val) for c_val in est_and_ci_B]
outcome_data_A.extend(est_and_ci_A)
outcome_data_B.extend(est_and_ci_B)
study_a_val = outcome_data_A[data_index]
study_b_val = outcome_data_B[data_index]
return self._meta_cmp_wrapper(study_a, study_b, study_a_val, study_b_val, reverse)
def helper_set_current_effect(ma_unit, txt_boxes, current_effect, group_str, data_type, mult=None):
'''Fills in text boxes on calculator forms with data from ma unit.
I noticed all 3 set_current_effect functions in the 3 calculators are
nearly identical so it makes sense to share the similiar parts'''
if mult is None:
raise ValueError("mult must be specified")
if data_type == "binary":
conv_to_disp_scale = lambda x: meta_py_r.binary_convert_scale(x, current_effect, convert_to="display.scale")
elif data_type == "continuous":
conv_to_disp_scale = lambda x: meta_py_r.continuous_convert_scale(x, current_effect, convert_to="display.scale")
elif data_type == "diagnostic":
conv_to_disp_scale = lambda x: meta_py_r.diagnostic_convert_scale(x, current_effect, convert_to="display.scale")
else:
raise Exception("data_type unrecognized")
effect_tbox, lower_tbox, upper_tbox = [txt_boxes[box_name] for box_name in ("effect","lower","upper")]
(est,lower,upper) = ma_unit.get_effect_and_ci(current_effect, group_str, mult)
(d_est,d_lower,d_upper) = [conv_to_disp_scale(x) for x in (est,lower,upper)]
for val, txt_box in zip((d_est,d_lower,d_upper),
[effect_tbox, lower_tbox, upper_tbox]):
txt_box.blockSignals(True)
if val is not None:
txt_box.setText(QString("%s" % round(val, CALC_NUM_DIGITS)))
else:
txt_box.setText(QString(""))
txt_box.blockSignals(False)
def f(study_a, study_b):
ma_unit_A = study_a.get_ma_unit(outcome_name, follow_up)
ma_unit_B = study_b.get_ma_unit(outcome_name, follow_up)
if outcome_type is BINARY:
to_display_scale = lambda x: meta_py_r.binary_convert_scale(
x, current_effect, convert_to="display.scale")
elif outcome_type is CONTINUOUS:
to_display_scale = lambda x: meta_py_r.continuous_convert_scale(
x, current_effect, convert_to="display.scale")
elif outcome_type is DIAGNOSTIC:
to_display_scale = lambda x: meta_py_r.diagnostic_convert_scale(
x, current_effect, convert_to="display.scale")
if outcome_type in (BINARY, CONTINUOUS):
outcome_data_A = ma_unit_A.get_effect_and_ci(
current_effect, group_str, mult)
outcome_data_B = ma_unit_B.get_effect_and_ci(
current_effect, group_str, mult)
outcome_data_A = [
to_display_scale(c_val) for c_val in outcome_data_A
]
outcome_data_B = [
to_display_scale(c_val) for c_val in outcome_data_B
]
elif outcome_type == DIAGNOSTIC:
# /\/\/\/\
(outcome_data_A, outcome_data_B) = ([], [])
# |
# \_____/
for diag_metric in [
"Sens", "Spec"
]: # this order corresponds to the order displayed on the spreadsheet
est_and_ci_A = ma_unit_A.get_effect_and_ci(
diag_metric, group_str, mult)
est_and_ci_B = ma_unit_B.get_effect_and_ci(
diag_metric, group_str, mult)
est_and_ci_A = [
to_display_scale(c_val) for c_val in est_and_ci_A
]
est_and_ci_B = [
to_display_scale(c_val) for c_val in est_and_ci_B
]
outcome_data_A.extend(est_and_ci_A)
outcome_data_B.extend(est_and_ci_B)
study_a_val = outcome_data_A[data_index]
study_b_val = outcome_data_B[data_index]
return self._meta_cmp_wrapper(study_a, study_b, study_a_val,
study_b_val, reverse)
def reset_conf_level(self):
print("Re-scaling est, low, high to standard confidence level")
old_effect_and_ci = self.ma_unit.get_effect_and_ci(self.cur_effect, self.group_str)
argument_d = {"est" : old_effect_and_ci[0],
"low" : old_effect_and_ci[1],
"high" : old_effect_and_ci[2],
"orig.conf.level": self.CI_spinbox.value(),
"target.conf.level": meta_py_r.get_global_conf_level()}
res = meta_py_r.rescale_effect_and_ci_conf_level(argument_d)
if "FAIL" in res:
print("Could not reset confidence level")
return
res["display_est"] = meta_py_r.diagnostic_convert_scale(res["est"], self.cur_effect, convert_to="display.scale")
res["display_low"] = meta_py_r.diagnostic_convert_scale(res["low"], self.cur_effect, convert_to="display.scale")
res["display_high"] = meta_py_r.diagnostic_convert_scale(res["high"], self.cur_effect, convert_to="display.scale")
# Save results in ma_unit
self.ma_unit.set_effect_and_ci(self.cur_effect, self.group_str, res["est"],res["low"],res["high"])
self.ma_unit.set_display_effect_and_ci(self.cur_effect, self.group_str, res["display_est"],res["display_low"],res["display_high"])
def helper_set_current_effect(ma_unit,
txt_boxes,
current_effect,
group_str,
data_type,
mult=None):
'''Fills in text boxes on calculator forms with data from ma unit.
I noticed all 3 set_current_effect functions in the 3 calculators are
nearly identical so it makes sense to share the similiar parts'''
if mult is None:
raise ValueError("mult must be specified")
if data_type == "binary":
conv_to_disp_scale = lambda x: meta_py_r.binary_convert_scale(
x, current_effect, convert_to="display.scale")
elif data_type == "continuous":
conv_to_disp_scale = lambda x: meta_py_r.continuous_convert_scale(
x, current_effect, convert_to="display.scale")
elif data_type == "diagnostic":
conv_to_disp_scale = lambda x: meta_py_r.diagnostic_convert_scale(
x, current_effect, convert_to="display.scale")
else:
raise Exception("data_type unrecognized")
effect_tbox, lower_tbox, upper_tbox = [
txt_boxes[box_name] for box_name in ("effect", "lower", "upper")
]
(est, lower, upper) = ma_unit.get_effect_and_ci(current_effect, group_str,
mult)
(d_est, d_lower,
d_upper) = [conv_to_disp_scale(x) for x in (est, lower, upper)]
for val, txt_box in zip((d_est, d_lower, d_upper),
[effect_tbox, lower_tbox, upper_tbox]):
txt_box.blockSignals(True)
if val is not None:
txt_box.setText(QString("%s" % round(val, CALC_NUM_DIGITS)))
else:
txt_box.setText(QString(""))
txt_box.blockSignals(False)
