def test_config_path_helper(self):
# if it's none, return the default
del os.environ['herd_config_path']
self.assertEqual(config_path(), os.path.expanduser("~/.herdconfig"))
# if it's something return that
os.environ['herd_config_path'] = "/mock/path/.herdconfig"
self.assertEqual(config_path(), "/mock/path/.herdconfig")
def setconfig(section, key, value):
""" set the key to the value in .herdconfig """
conf = ConfigParser()
with open(config_path(), 'r') as configfile:
conf.readfp(configfile)
conf.add_section(section)
conf.set(section, key, value)
with open(config_path(), 'w') as configfile:
conf.write(configfile)
def get_past_diagnosis(patient_id):
config = config_path()
patients_files = json.loads(open(config.patient_dict).read())
history = patients_files[patient_id]
return (history)
def test_init():
path = cfg.config_path()
run = cfg.config_run()
param = cfg.config_param()
test_param = cfg.param_test()
csv_path = os.path.join(path.data_dir_top, test_param.csv_outs)
if os.path.exists(csv_path):
print("%s is exist!" % (csv_path))
else:
os.makedirs(csv_path)
print("creat the dir %s!" % (csv_path))
event_path = os.path.join(path.data_dir_top, path.event_top_dir,
test_param.event_dir)
if os.path.exists(event_path):
print("%s is exist!" % (event_path))
else:
os.makedirs(event_path)
print("creat the dir %s!" % (event_path))
FLAGS.subset = str(run.mode)
## new param mode
FLAGS.test_dir = str(event_path)
FLAGS.checkpoint_dir = str(
os.path.join(path.data_dir_top, test_param.model_dir))
FLAGS.data_dir = str(
os.path.join(path.data_dir_top, test_param.tfrecord_dir))
FLAGS.csv_dir = csv_path
FLAGS.input_queue_memory_factor = test_param.input_queue_memory_factor
FLAGS.num_examples = test_param.num_example
FLAGS.batch_size = test_param.batch_size
FLAGS.run_once = test_param.run_once
FLAGS.test_gpu_id = test_param.gpu_id
return 0
def main():
path = cfg.config_path()
if main_init() != 0:
return 1
tmpl_param = get_tmpl_param()
#tmpl land_mask computer
raw_dir = os.path.join(path.top_dir, path.raw_dir)
list_dir = os.listdir(raw_dir)
for name in list_dir:
name_path = os.path.join(path.top_dir, path.raw_dir, name)
img_dat = mpimg.imread(name_path)
hsd_dat = hsd_t.rgb2hsd(img_dat)
he_label = pixc.patch_pixel_cluster(img_dat, 3)
#ing_info include img_param and img_lm_label
img_info = get_param(hsd_dat, he_label)
#change mean and angle
img_mean = img_info[0]
img_angle = img_info[1]
an_chg_hsd_h, an_chg_hsd_e = mean_angel_perprocess(
hsd_dat, he_label, img_mean, img_angle)
#get land mask ,and get img landmask label
lm = exfeat.get_all_landmask(an_chg_hsd_h, an_chg_hsd_e, he_label)
lm_label = lm_lab.get_all_lm_label()
hsd_out = transform.transform_to_tmpl(hsd_dat, he_label, img_info,
tmpl_param)
save_hsd_img(hsd_out, name)
return 0
def save_hsd_img(hsd, name):
path = cfg.config_path()
outs_path = os.path.join(path.top_dir, path.hsd_dir, name)
img_dat = hsd_t.hsd2rgb(hsd)
mpimg.imsave(outs_path, img_dat)
return 0
def get_tmpl_param():
path = cfg.config_path()
tmpl_path = os.path.join(path.top_dir, path.tmpl_name)
img_dat = mpimg.imread(tmpl_path)
hsd_dat = hsd_t.rgb2hsd(img_dat)
he_label = pixc.patch_pixel_cluster(hsd_dat)
return get_param(hsd_dat, he_label)
def get_current_goals():
config = config_path()
goals_file = json.loads(open(config.goals_dict).read())
goals = goals_file[config.patient_id]
if len(goals) == 0:
return (('example: Hba1c level < 7;'))
else:
return (('\n'.join(goals) + ';'))
def eval_init():
path = cfg.config_path()
#run = cfg.config_run()
param = cfg.config_param()
eavl_param = cfg.param_eval()
FLAGS.num_gpus = param.gpu_num
FLAGS.eval_dir = str(
os.path.join(path.data_dir_top, path.event_top_dir,
eavl_param.event_dir))
FLAGS.data_dir = str(
os.path.join(path.data_dir_top, eavl_param.tfrecord_dir))
FLAGS.checkpoint_dir = str(
os.path.join(path.data_dir_top, eavl_param.model_dir))
FLAGS.input_queue_memory_factor = eavl_param.input_queue_memory_factor
FLAGS.num_examples = eavl_param.num_example
FLAGS.batch_size = eavl_param.batch_size
FLAGS.subset = eavl_param.mode_name
FLAGS.run_once = eavl_param.run_once
return 0
def train_init():
path = cfg.config_path()
run = cfg.config_run()
param = cfg.config_param()
save_models_path = os.path.join(path.data_dir_top, path.save_models_dir)
if os.path.exists(save_models_path):
print("save models dir exist!")
else:
print("creat save models dir!")
os.makedirs(save_models_path)
FLAGS.train_dir = str(save_models_path)
FLAGS.data_dir = str(os.path.join(path.data_dir_top, path.data_dir))
FLAGS.fine_tune = param.fine_tune
FLAGS.initial_learning_rate = param.init_learning_rate
FLAGS.input_queue_memory_factor = param.input_queue_memory_factor
FLAGS.checkpoint_dir = str(os.path.join(path.data_dir_top,
path.models_dir))
FLAGS.method_name = str(run.method_name)
#FLAGS.pretrained_model_checkpoint_path = str(os.path.join(path.data_dir_top,path.models_dir,path.model_checkpoint))
FLAGS.batch_size = param.train_batch_size
FLAGS.num_gpus = param.gpu_num
FLAGS.max_steps = param.max_step
FLAGS.subset = str(run.mode)
#!/usr/bin/env python3
import json
import os
import numpy as np
import pandas as pd
import networkx as nx
from config import config_path
config = config_path()
def graph_sugg():
transcripts_dir = 'Transcripts'
with open(config.transcript_D, 'r') as f:
current_diagnoses = f.read()
def neighbours_suggestions(graph, current_diagnoses, past_diagnoses):
# Neighbours of the current diagnosis
try:
first = sum(
[list(nx.all_neighbors(graph, v)) for v in current_diagnoses],
[])
except nx.exception.NetworkXError:
first = []
# give weight 1
first = list(zip(first, np.repeat(1, len(first))))
# Neighbours of the past diagnoses
try:
second = sum(
